<<

NEWSLETTER | The American Philosophical Association

Philosophy and Computers

SPRING 2015 VOLUME 14 | NUMBER 2

FROM THE GUEST EDITOR John P. Sullins

NOTES FROM THE COMMUNITY ON PAT SUPPES

ARTICLES

Patrick Suppes Patrick Suppes Autobiography

Luciano Floridi Singularitarians, AItheists, and Why the Problem with Artificial is H.A.L. (Humanity At Large), not HAL

Peter Boltuc First-Person as Hardware

D. E. Wittkower Social Media and the Organization Man

Niklas Toivakainen The Moral Roots of Conceptual Confusion in Research

Xiaohong Wang, Jian Wang, Kun Zhao, and Chaolin Wang Increase or Decrease of Entropy: To Construct a More Universal Macroethics (A Discussion of Luciano Floridi’s The of Information)

VOLUME 14 | NUMBER 2 SPRING 2015

© 2015 BY THE AMERICAN PHILOSOPHICAL ASSOCIATION ISSN 2155-9708 APA NEWSLETTER ON and Computers

JOHN P. SULLINS, GUEST EDITOR VOLUME 14 | NUMBER 2 | SPRING 2015

but here we wish to celebrate his accomplishments in the FROM THE GUEST EDITOR fields of philosophy and computing one last time.

John P. Sullins To accomplish that goal I have compiled some interesting SONOMA STATE UNIVERSITY pieces from an autobiography that Pat wrote some years ago but that he added to a bit for an event held in his honor November 17, 2014, marked the end of an inspiring at Stanford. In this document he explains his motivations career. On that day Patrick Suppes died quietly at the and accomplishments in various fields of study that are age of ninety-two in his house on the Stanford Campus, of interest to our community. In that section you will see which had been his home both physically and intellectually just how ambitious Pat was in the world of computer since 1950. At the time of his death he was the Lucie education, particularly using the computer as a tool for Stern Professor of Philosophy, Emeritus and a member of philosophy education. While few would argue the the departments of Statistics and Psychology and of the of computer and online education when it comes to the Graduate School of Education. Pat was the first recipient of formal topics in philosophy, such as basic or aspects the Barwise Prize, which is awarded by the APA Committee of critical thinking, I think many would argue that these on Philosophy and Computers. A more fitting for tools are grossly inappropriate for developing the Socratic the prize would have been hard to find given that Pat has symposium style of instruction have used done significant research, teaching, and publishing on a since the beginning of the profession many millennia vast array of subjects, many of which crossed over into the ago. Even from the early days of personal computing, Pat areas that we now call the philosophy of computers and saw things differently. He was convinced that something information. This, of course, was neither his only nor even like an AI tutor could be developed that could act as a his most significant award. In 1990 he was awarded the Socratic tutor to a student, and take them through a rich, National Medal of Science for his work in the measurement discussion based, educational experience that introduced of subjective probability and utility in uncertain situations, them to not only the facts of philosophy but to the the development and testing of general learning theory, method of philosophical inquiry as well. As you will see, the and syntax of natural , and the use he also imagined the “flipped” classroom many decades of interactive computer programs for instruction, any one before that term came into vogue when he suggests that of which could have been sufficient for an entire career. the computer is best used in philosophy to prepare the This issue of the newsletter is dedicated to his memory student on facts and reading so that they come prepared and towards that end I have collected some interesting to enter the classroom to enter into vigorous philosophical material. discussions. Pat was also way ahead of the game when it comes to experimental philosophy. He considered While I only knew Pat from meeting him at various himself an empiricist first and foremost, and he believed in conferences and events, I was struck by his generosity testing his through experimentation. When Michael of spirit and the way he was undaunted by age. He never Friedman from the Suppes Center for the History and really retired from his positions at Stanford, he just went at Stanford and John Markoff from on to find other ways to contribute and work within the the New York Times were working on writing obituaries for academic community there and was still active with some Pat, John found a news clip from the Times archive from research and teaching in the spring of 2014. The last time 1966 describing one of Pat’s first large-scale experiments I saw him was at a conference at Stanford celebrating in deploying computer education to first grade students the many significant achievements that they have had in in an East Palo Alto school.1 East Palo Alto is very different the philosophy of science. Pat was there and in his glory from Palo Alto, its more affluent neighbor, but somehow Pat since many, if not most of those achievements were things convinced IBM to put the computer with sixteen consoles, that he had contributed substantially to. He was right in costing $450,000 dollars in 1966 money, into a school that the thick of the discussion the entire event. The lasting served underprivileged and at-risk students. Pat wanted legacy I will take from him is as a role model for the joyous the promise of computer education to cross all the social, dedication to the life of the and the many pleasures political, race, and economic boundaries. True to his vision, that can bring right to one’s last days. It is sad to lose him; this machine, the IBM 1500, made use of verbal inputs and I think he had at least another fifty years of ideas left outputs through which “the child gets spoken commands, in him but he has bequeathed us much that we can use and suggestions and encouragement.”2 As John Markoff notes build on. There have been a number of very fine obituaries in the obituary he wrote for the Times, Pat wanted everyone written for Pat, and we have links to those later in the issue, to have a computerized tutor that could personally attend APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

to them as they learned any and every . Of course, this would be a very ambitious claim to make now and was NOTES FROM OUR COMMUNITY even more in the 1960s when he first started developing to make it happen. ON PAT SUPPES

One of Pat’s most enduring legacies will be his many positive interactions with the he worked with. To address that I also solicited some remembrances of Pat from our various members, and in that section of the newsletter you will find a wonderful anecdote from Marvin Croy that paints an amusing portrait of Pat as well as illustrates his desire to help younger scholars.

In addition, we will dive into a brewing controversy. Recently, wrote a scathing critique of Luciano Floridi’s book, The Fourth Revolution in the New York Review of Books. Floridi was able to write a short reply in the pages of the review, but I it would be interesting to give him as much space as he wanted to make a more thorough reply and we have that in this issue. Hopefully it will spark an interesting discussion in the pages of this newsletter. Floridi’s work continues to gain attention across the globe, As IACAP program chair in 2002 I had the honor of and we have a good discussion of his book The Ethics introducing him with his presentation “A Retrospective on of Information from Professor Xiaohong et al. from Xi’an Instructional Computing” which was excellent. We shared Jiaotong University, P. R. China. many moments together during the meeting which I shall cherish always. Building on the philosophy of AI theme for this issue, we – Ron Barnette have two good articles that make challenging claims within the philosophical discussion on AI. Pete Boltuc makes an Pat Suppes had an eye for the future, and helped to lead interesting case for the that first-person consciousness us there. In 1967, he published “On Using Computers fits with a hardware analogy better than the idea that it is to Individualize Instruction,” The Computer in American similar to a software process. In this way he hopes to find Education (1967): 11–24. It is startling that 47 years later, middle ground between reductivist and non-reductivist that paper still has something interesting to say about the arguments. Late in the issue Niklas Toivakainen makes use of computing in education. the case that there is a deep conceptual confusion in the philosophical discussions of AI that may be preventing – Keith W. Miller, Orthwein Endowed Professor for meaningful dialog. Rounding out the issue, we have a Lifelong Learning in the Sciences, University of paper from D. E. Wittkower that explores some of the Missouri–St. Louis ethical impacts of social media from a new point of view. Please, do not forget the work done jointly with Mario We accept submissions regularly for publication in the Zanotti, see e.g.: Foundations of Probability with newsletter. This is a good forum for following up on debates Applications: Selected Papers 1974–1995, Patrick Suppes or making comments on discussions that might not fit well and Mario Zanotti, Cambridge University Press, 1996. in other journal formats. We also like to see position papers and reviews that can spark productive conversations. – Stefano Cerri, Montpellier Laboratory of Informatics, Please send anything you want to see in the newsletter in Robotics, and Microelectronics (LIRRM) and French a timely manner. Computers and philosophy is a topic that National Center for Scientific Research (CNRS) moves very quickly, and philosophers need to play a role in the development and analysis of these world-changing Had I never met Pat Suppes, there is no question that I technologies. would not have spent the last thirty plus years developing and researching instructional computer programs for NOTES teaching logic. In 1978, I was a grad student at Florida 1. Thomas O’Toole, “A Teacher Is Hired on Coast: 170 First State studying philosophy of science. I traveled to Graders to Begin Daily Use of a Computer,” New York Times (1923– Stanford for a week to help investigate the possibility of current file); April 4, 1966; ProQuest Historical Newspapers: the transporting the Stanford logic program to FSU. What I New York Times (1851–2010), 35. witnessed in action within Suppes’s shop was what we now 2. Ibid. call interdisciplinary applied research. This is common in science today, but then most computer-assisted instruction projects for teaching logic were focused on successfully delivering drill and practice lessons. Suppes’s own system for constructing deductive proofs went far beyond this, and was the centerpiece of a complete course in logic. Moreover, Suppes directed a team approach, working

PAGE 2 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

closely with psychologists, computer scientists, and even speech technologists and health-care providers when it ARTICLES came to teaching deaf or blind students. Patrick Suppes Autobiography Now, Pat was interested in just about every topic, and he made significant contributions to many. This came home Patrick Suppes in an amusing way when I went to his office for a chat. It turned out that literally every square inch of his desk The following is a document that was prepared for use at was piled high with books, so high that when I sat down two events held at Stanford University where Pat Suppes was I could not see him across the desk! I tried not to look honored for his many years of dedication to the intellectual surprised and slid sideways to intersect with a valley in life at that university. It appears to be an abridged version the mountains of tomes, where we had eye contact and a of a document titled, “Intellectual Autobiography (Written wide-ranging conversation. (I laugh now, but at the time it in 1978),”1 which has been edited and added to at some felt very bizarre!) I am grateful to have experienced Pat’s later date by Patrick Suppes. What follows is a transcription energetic and optimistic style of engaging life, then and of that document.2 over the years. FOUNDATIONS OF PHYSICS Pat’s project made a lasting impression on me, and the My doctoral dissertation lay within the philosophy of most important influence concerned the nature of data physics. In particular, I studied the problem of action at collection. Computer-assisted instruction programs a distance as it had occurred in 17th- and 18th-century were routinely designed to collect information for physics and philosophy, especially in the writings of immediate feedback and grading. But Pat took this to a Descartes, Newton, Boscovich, and Kant. The final new level, carefully collecting data to address questions chapter dealt with the problem in the special theory of concerning interface , aspects of the subject matter relativity. Working on it strengthened my earlier desire to that students found most demanding, and program features give an axiomatic formulation of classical mechanics in proved to be most effective, all of which generated a new the spirit of modern rather than “physical” round of development questions. In , this makes axiomatizations common in physics. Serious joint work every instructional project also a research on this project began soon after my arrival at Stanford, in project, a key point when developers come up for tenure collaboration with J. C. C. McKinsey, and is represented and funding decisions. Today we call this the scholarship of in four papers we wrote on the foundations of physics teaching and learning. In 1978, I came away calling it one prior to McKinsey’s death in 1953 (1953a, 1953b, I953c hell of an adventure. Thanks, Pat!! also with A. C. Sugar, and 1955b). Shortly thereafter I wrote with Herman Rubin a similar paper (1954c) on the – Marvin Croy, Complex Systems Institute, axiomatic foundations of relativistic particle mechanics. It University of North Carolina–Charlotte is a long and very complicated piece of work that has not been read, I suspect, by very many people.

LINKS TO OBITUARIES FOR PATRICK SUPPES QUANTUM MECHANICS Most of the effort that I have put in on the foundations Stanford Philosophy Obituary of physics since 1960 has been devoted to quantum Stanford News Service mechanics, and this continues to be a current active New York Times Obituary intellectual interest. Almost everything that I have written Los Angeles Times Obituary about quantum mechanics has been intertwined with Stanford Daily article questions related to the foundations of probability, especially as to how probabilistic are used in quantum mechanics. My first paper on the subject (1961c) was concerned with the absence of a joint distribution of position and momentum in many standard cases. I shall not enter into the technical details of the argument here, but I do want to convey the basic philosophical point that I continue to find puzzle of quantum mechanics. Not the move away from classical determinism, but the ways in which the standard versions seem to lie outside the almost universal methodology of modern probability theory and mathematical statistics. For me it is in this arena that the real puzzles of quantum mechanics are to be found. I am philosophically willing to violate classical physical without too many qualms, but when Pat Suppes accepting the first APA Barwise Prize it comes to moving away from the broad conceptual and in Philosophy and Computing in 2002. Robert formal framework of modern probability theory I am at Cavalier from Carnegie Mellon presented the prize and sitting beside Suppes is Richard Scheines once uneasy. My historical view of the situation is that if (now Dean of Carnegie Mellon’s Dietrich College of probability theory had been developed to anything like its Humanities and Social Sciences). current sophisticated state at the time the basic work on

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 3 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

quantum mechanics was done in the twenties, then a very Shortly after this I began to think more generally about different sort of theory would have been formulated. the foundational aspects of theories of measurement and was fortunate to have as a collaborator the logician It is worth recording a couple of impressions about this and mathematician , who was at that time a because they indicate the kind of changes that can take graduate student in mathematics. (Scott is also one of the place in one’s attitudes as the years go by. Initially I was Berkeley-Stanford persons from whom I learned a great much impressed by the mathematical formulation of deal, beginning when he was an undergraduate in a course foundations given by Von Neumann in his classical work on the philosophy of science I taught at Berkeley in 1952, and, later, by Mackey (1963), whose book has also become along with Richard Montague. What a pair to have in such classical in its own way. No doubt I was originally struck a course!). Scott and I tried to give a general framework by the mathematical clarity and sophistication of this for theories of measurement and to obtain some specific work, but in later years I have become dissatisfied with results about axiomatization. This article was published the unsatisfactory conceptual basis from a probabilistic in 1958, a year or so after it was written. The framework standpoint of the way in which the theory is formulated. that Scott and I set up has, I think, been of use in the I shall give here just two examples to indicate the nature literature, and probably the article with him has been the of my conceptual dissatisfaction. Von Neumann stresses most important article in the theory of measurement that that we can take the expectation of the sum of any two I have written, although the chapter in the Handbook of operators, even though they are conjugate, that is, do not Mathematical Psychology, written with J. L. Zinnes and commute. But once this is said, the natural question is to published in 1963, has perhaps been more influential, ask about the underlying probability space that justifies the especially in psychology. exact probabilistic of the expectation. A similar question arises with respect to Mackey’s treatment. Mackey DECISION THEORY takes as fundamental the of the probability that a It is not easy to disentangle measurement theory and measurement in a given state of an observable will lead to decision theory because the measurement of subjective a given value. This seems innocent enough, but when the probability and utility has been such a central part of fundamental postulates of the theory are stated in these decision theory. The separation that I make will therefore terms, what seems missing from what one would expect be somewhat arbitrary. My really serious interest in in a standard causal physical theory is any clarity about the psychology began with experimental research on decision relation between observables. The axioms he gives would theory in collaboration with my philosophical colleague seem to concentrate too deeply on the relatively simple Donald Davidson and a graduate student in psychology of the probability of a given measurement at that time, Sidney Siegel. Davidson and I had begun on a given observable and not enough on the causal collaborative work with McKinsey in 1953 on the theory dependencies between observables. (It is important to of value and also on utility theory. We continued this work remember that I am not really making a technical argument after McKinsey’s death, and it is reflected in Davidson, here but trying to give the back of arguments that McKinsey, and Suppes (1955a) and in the joint article I think can be formalized.) with Davidson (1956b) on the finitistic axiomatization of subjective probability and utility, already mentioned. THEORY OF MEASUREMENT The article on the measurement of utility based on utility In my first published article (1951a) I gave a set of differences, with Muriel Winet, was also part of this effort. independent axioms for extensive quantities in the tradition of earlier work by Holder and Nagel. My contribution was Sometime during the year 1954, Davidson and I undertook, primarily to weaken the assumptions of Holder axioms with the collaboration of Siegel, an experimental investigation and also to prove that both the axioms and the concepts of the measurement of utility and subjective probability. used were independent. Looking around for other topics Our objective was to provide an explicit methodology for in measurement, and returning to the earlier interest in separating the measurement of the two and at the same the theory of games and utility theory, it soon became time to obtain conceptually interesting results about the apparent that there were more outstanding problems of character of individual utility and probability functions. measurement in psychology than in physics. One of my This was my first experimental work and consequently in first efforts in this direction was a joint article with my a genuine sense my first real introduction to psychology. student Muriel Winet (1955d). We gave an axiomatization The earlier papers on the foundations of decision theory of utility based on the notion of utility differences. The concerned with formal problems of measurement were a idea of considering such utility differences is a very old natural and simple extension of my work in the axiomatic one in the literature, but an explicit and adequate set of foundations of physics. Undertaking experimental work axioms had not previously appeared. In 1956 I published was quite another matter. I can still remember our many two other articles which fell between decision theory and quandaries in deciding how to begin, and seeking the measurement theory. One was on the role of subjective advice of several people, especially our colleagues in the probability and utility in decision making. In this article Department of Psychology at Stanford. (1956b) I used the results of the joint work with Winet to provide an axiomatization alternative to that given by I continued a program of experimentation in decision Savage in his book Foundations of Statistics (1954). And theory as exemplified in the joint work with Halsey Royden in the second article, my colleague Donald Davidson and I and Karol Walsh (1959i) and the development of a nonlinear gave a finitistic axiomatization of subjective probability and model for the experimental measurement of utility with utility (1956c). Walsh (1959j).

PAGE 4 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

DISTRIBUTIVE CAUSALITY For a variety of , the literature on decision theory Because my own approach to causality is probabilistic in has been interwined [sic] with the literature on social character, I have included it in this section. It is hard to think choice theory for a very long period, but the focus of the of a philosophical topic that has received more attention two literatures is rather different and I have certainly had historically than that of causality. It has already become more to say about decision theory than about the normative clear to me that what I have had to say (1970a) has got problems of social choice or distributive justice. To a large to be extended, revised, and deepened, in order to meet extent, this is an accident of where I have happened to objections that have been made by other people and to have had some ideas to develop and not a matter of a priori account for a variety of phenomena that I did not consider choice. I have published two papers on distributive justice in any detail. Causality is one of those concepts that plays a (1966, 1977a). The main results about justice in the first major role in a variety of scientific disciplines and that can be one, which were stated only for two persons, were nicely clarified and enriched by extensive . generalized by Amartya Sen (1970). The other paper, which On some subjects of a probabilistic kind I find it hard to was just recently published, looks for arguments to defend imagine how I, or another , could improve in unequal distributions of income. I am as suspicious of a substantial way on what has been said with clarity and simplistic arguments that lead to a uniform distribution of precision by probabilists and statisticians—the concept of income as I am of the use of the principle of indifference in a stochastic process is a good example. This is not true of the theory of beliefs to justify a uniform prior distribution. the concept of causality. A good many statisticians use the The arguments are too simple and practices in the real concept in various ways in their research and writing, and world are too different. A classical economic argument the concept has been a matter of controversy both in the to justify inequality of income is productivity, but in all physical sciences and in the social sciences over the past societies and economic subgroups throughout the world several decades. There is a major place in these discussions differences in income cannot be justified purely by claims for philosophical analyses of causality that join issue firmly about productivity. Perhaps the most universal principle and squarely with this extensive scientific literature. also at work is one of seniority. Given the ubiquitous character of the preferential arising from seniority SET-THEORETICAL METHODS in the form of income and other rewards, it is surprising I do not think of set-theoretical methods as providing any how little conceptual effort seems to have been addressed absolute kind of clarity or certainty of results independent to the formulation of principles that justify such universal of this particular point in the history of such matters. practices. They constitute a powerful instrument that permits us to communicate in a reasonably objective way the structure FOUNDATIONS OF PROBABILITY of important and complicated theories. In a broad spirit The ancient Greek view was that time is cyclic rather than they represent nothing really new; the axiomatic viewpoint linear in character. I hold the same view about my own that underlies them was developed to a sophisticated pattern of research. One of my more recent articles (1974g) degree in Hellenistic times. Explicit use of such methods is concerned with approximations yielding upper and provides a satisfactory analysis of many questions that lower probabilities in the measurement of partial belief. were in the past left vaguer than they need to be. A good The formal theory of such upper and lower probabilities example would be their use in the theory of measurement in qualitative terms is very similar to the framework for to establish appropriate isomorphic relations between extensive quantities developed in my first paper in 1951. qualitative empirical structures and numerical structures. In retrospect, it is hard to understand why I did not see the simple qualitative analysis given in the 1974 paper at the CONCLUSION time I posed a rather similar problem in the 1951 paper. [Document ends here] The intuitive idea is completely simple and straightforward: A set of “perfect” standard scales is introduced, and then * * * the measurement of any other event or object (event in the case of probability, object in the case of mass) is The document above omits quite a bit of the work that Pat made using standard scales just as we do in the ordinary did up until the late seventies, and given the interest of use of an equal-arm balance. This is not the only occasion the readers of this newsletter we will excerpt the sections in which I have either not seen an obvious and simple on Education and Computers and Computer-assisted approach to a subject until years later, or have in fact instruction from the original document. missed it entirely until it was done by someone else. . . . Recently we have found correspondingly simple necessary * * * and sufficient qualitative axioms for conditional probability. The qualitative formulations of this theory beginning with EDUCATION AND COMPUTERS the early work of B. O. Koopman (1940a, I940b) have been In the section on mathematical concept formation in especially complex. We have been able drastically to children I mentioned the beginning of my interests in simplify the axioms by using not only extended indicator education in 1956 when my oldest child, Patricia, entered functions, but the restriction of such functions to a given kindergarten. I cited there the work in primary-school event to express conditionalization. In the ordinary logic geometry. An effort, also noted but briefly, that was of events, when we have a conditional probability P(A|B), much more sustained on my part was work in the basic there is no conditional event A|B, and thus it is not possible elementary-school mathematics curriculum. This occupied to define operations on conditional or restricted events. a fair portion of my time between about 1956 and the middle

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 5 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

of the sixties and led to publication of a basic elementary- my own publications, including the two books listed above, school mathematics textbook series, Sets and Numbers, and a comparable body of publications has issued from the which was one of the more radical of the “new math” work of Atkinson and his colleagues on initial reading. My efforts. Unlike many of my colleagues in mathematics and own experience has been that even a subject as relatively science who became interested in school curriculum after simple as elementary-school mathematics is of unbounded Sputnik, I had a genuine interest in the psychological and complexity in terms of the underlying empirical aspects of learning and a traditional interest in psychological theory of learning and performance. Over knowing what had been done before. the past several years I have found myself moving away from the kind of framework that is provided by stimulus When I began working on the foundations of physics sampling theory and that has been so attractive to me after graduate school, I was shocked at the absence of for so many years. The new ideas are more cognitive in what I would call traditional scholarship in the papers of character and organized around the concept of procedures philosophers like Reichenbach that I read, or even more or programs as exemplified, for instance, in a simple of physicists who turned to philosophical matters such as register machine, that is, a simple idealized computer with Bridgman and Campbell. There was little or no effort to a certain number of registers and a small, fixed number know anything about the previous serious work in the field. of instructions (1973c). I think that the ideas of stimulus I found this same attitude to be true of my colleagues from sampling theory still have importance in terms of learning, the sciences who became interested in education. They even in the context of such procedures or programs, but had no desire to know anything about prior scholarship in certainly there is a shift in conceptual interest characteristic education. not only of my own work but also of that of a great many psychologists originally devoted to learning. I found I had a real taste for the concrete kinds of questions that arise in organizing a large-scale curriculum activity. I One of my initial interests in computer-assisted instruction shall not attempt to list all the aspects of this work here, but was the teaching of logic at the elementary-school level since, beginning in the mid-fifties, I have written a large and subsequently at the college level. Once complexity number of research papers concerned with how students of this level is reached, psychological theory is in a more learn elementary mathematics and I have had a fairly large difficult spot in terms of providing appropriate conceptual number of students from education or psychology write tools for the analysis of student behavior. Currently my dissertations in this area. Most of the work in the last work in computer-assisted instruction is almost entirely decade or so has been within the context of computer- devoted to university-level courses, and we are struggling assisted instruction, to which I now turn. to understand how to analyze data from the sorts of proofs or logical derivations students give in the first logic course COMPUTER-ASSISTED INSTRUCTION or in the course in axiomatic that follows it. In the fall of 1962, on the basis of conversations with Lloyd Morrisett, Richard Atkinson and I submitted a proposal to Although there are many questions about the psychology the Carnegie Corporation of New York for the construction of of learning and performance in elementary-school a computer-based laboratory dedicated to the investigation mathematics that I do not understand, still I feel that I of learning and teaching. The proposal was funded in have a relatively deep conceptual grasp of what is going January 1963 and the laboratory began operation in the on and how to think about what students do in acquiring latter part of that year as computing equipment that was elementary mathematical skills. This is not at all the case ordered earlier in the year arrived and was installed. The for skills of logical inference or mathematical inference, laboratory was initially under the direction of an executive as exemplified in the two college-level courses I have committee consisting of Atkinson, Estes, and me. In mentioned. We are still floundering about for the right addition, John McCarthy of the Department of Computer psychological framework in which to investigate the Science at Stanford played an important role in the design complete behavior of students in these computer-based and activation of the laboratory. In fact, the first computer courses. facilities were shared with McCarthy and his group. There are other psychological and educational aspects From a research standpoint, one of my own strong of the work in computer-assisted instruction that have motivations for becoming involved in computer-assisted attracted a good deal of my attention and that I think are instruction was the opportunity it presented of studying worth mentioning. Perhaps the most important is the extent subject-matter learning in the schools under conditions to which I have been drawn into the problems of evaluation approximating those that we ordinarily expect in a of student performance. I have ended up, in association psychological laboratory. The history of the first five years with my colleagues, in trying to conceive and test a number of this effort, through 1968, has been described in great of different models of evaluation, especially for the detail—probably too much detail for most readers—in two evaluation of performance in the basic skills of mathematics books (l968a, l972a) and in a large number of articles. I and reading in the elementary school. Again I will not try shall restrict myself here to a few general comments. to survey the various papers we have published except to mention the work that I think is probably intellectually To some extent those initial hopes have been realized of the most interesting and which is at the present time best obtaining school-learning data of the sort one expects reported in Suppes, Fletcher, and Zanotti (1976f), in which to get in the laboratory. Massive analyses of data on we introduce the concept of a student trajectory. The first elementary-school mathematics have been presented in point of the model is to derive from qualitative assumptions

PAGE 6 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

a differential equation for the motion of students through currently handle we will not move on to deeper solutions the course, initially the drill-and-practice supplementary in the future. Because I am able to teach all my own work in elementary mathematics given at computer undergraduate courses in a thoroughly computer-based terminals. The constants of integration of the differential environment, I now have, at the time of writing this essay, equation are individual constants of integration, varying the largest teaching load, in terms of number of courses, for individual students. On the basis of the estimation of any faculty member at Stanford. During each term I offer of the constants of integration we have been able to get ordinarily two undergraduate courses, one in logic and one remarkably good fits to individual trajectories through the in axiomatic set theory, both of which are wholly taught at curriculum. (A trajectory is a function of time, and the value computer terminals. In addition, I offer either one or two of the function is grade placement in the course at a given graduate seminars. As I have argued elsewhere on several time.) The development of these ideas has taken me back occasions, I foresee that computer technology will be one to ways of thinking about evaluation that are close to my of the few means by which we can continue to offer highly earlier work in the foundations of physics. technical and specialized courses that ordinarily draw low enrollment, because of the budgetary pressures that exist at Research on computer-assisted instruction has also all American universities and that will continue unremittingly provided the framework within which the large-scale throughout the remainder of this century. Before I am done empirical work on first-language learning in children has I hope to add other computer-based courses in relatively taken place. Without the sophisticated computer facilities specialized areas, such as the foundations of probability available to me at Stanford it would not have been possible and the foundations of measurement. The enrollment in to pursue these matters in such detail and on such a scale. one of these courses will ordinarily consist of no more than Even more essentially, the presence of a sophisticated five students. I shall be able to offer them only because I computer system in the Institute for Mathematical Studies can offer them simultaneously. My vision for the teaching in the Social Sciences has led to the computer-based of philosophy is that we should use the new technology approach to the problems of language learning and of computers to return to the standard of dialogue and performance mentioned earlier. One of our objectives intimate discourse that has such a long and honored for the future is to have a much more natural interaction tradition in philosophy. Using the technology appropriately between student and computer program in the computer- for prior preparation, students should come to seminars based courses we are concerned with. Out of these efforts ready to talk and argue. Lectures should become as passé I believe we shall also come to a deeper understanding of as the recitation methods of earlier times already have. not only how computer programs can best handle language but also how we do, in fact, handle it. (Part of this search for In 1967, when computer-assisted instruction was still a naturalness has led to intensive study of prosodic features very new educational technology, I organized with Richard of spoken speech and how to reproduce them in computer Atkinson and others a small company, Computer Curriculum hardware and software.) Corporation, to produce courses in the basic skills that are the main focus of elementary-school teaching. In retrospect I have not yet conveyed in any vivid sense the variety of it is now quite clear that we were ahead of our times and conceptual and technical problems of computer-assisted were quite lucky to survive the first five or six years. Since instruction that I have tried to deal with in collaboration with about 1973 the company has prospered, and I have enjoyed my colleagues since 1963. This is not the place to undertake very much my part in that development. I find that the kind a systematic review of these problems, most of which of carefully thought out and tough decisions required to have been dealt with extensively in other publications. I keep a small business going suits my temperament well. do, however, want to convey the view that the best work is yet to be done and will require solution of formidable I have not worked in education as a philosopher. I have intellectual problems. The central task is one well described published only one paper in the philosophy of education by long ago in ’s dialogue Phaedrus. Toward and read a second one, as yet unpublished, on the aims the end of this dialogue, Socrates emphasizes that the of education, at a bicentennial symposium. Until recently written word is but a pale image of the spoken; the highest I do not think I have had any interesting ideas about the form of intellectual discourse is to be found neither in philosophy of education but I am beginning to think about written works or prepared speeches but in the give and these matters more intensely and expect to have more to take of spoken arguments that are based on knowledge of say in the future. the truth. Until we have been able to reach the standard set by Socrates, we will not have solved the deepest problems * * * in the instructional use of computers. How far we shall be Above sections excerpted from Bogdan, R.J. (ed.), Patrick Suppes. able to go in having computer programs and accompanying Dordrecht, Holland: D. Reidel Publishing Company, 1979. Retrieved hardware that permit free and easy spoken interaction January 2015 from http://web.stanford.edu/~psuppes/autobio19.html. between the learner and the instructional program is not possible to forecast with any reasonable confidence, for NOTES we are too far from yet having solved simple problems of 1. R. J. Bogdan, ed., Patrick Suppes (Dordrecht, Holland: D. Reidel Publishing Company, 1979). Full text available as of 2015 at language recognition and understanding. http://web.stanford.edu/~psuppes/autobio1.html. This reprint is not meant to challenge the copyright of the original in any way. At the present time we are only able to teach well skills of 2. Many thanks to Dikran Karagueuzian, CSLI Publications Stanford, mathematics and language, but much can be done, and it Pat Suppes’s survivors, and the Pat Suppes Estate for their is my conviction that unless we tackle the problems we can gracious help in allowing us to print these materials.

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 7 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

earth to enslave us. Sometimes Singularitarianism is Singularitarians, AItheists, and Why the presented conditionally. This is shrewd because the then does follow from the if, and not merely in an Problem with Artificial Intelligence is quod libet sense: if some kind of were H.A.L. (Humanity At Large), not HAL to appear, then we would be in deep trouble. Correct. But this also holds true for the following conditional: if the Luciano Floridi Four Horsemen of the Apocalypse were to appear, then OXFORD INTERNET INSTITUTE, UNIVERSITY OF OXFORD we would be in even deeper trouble, trust me. Some other [email protected] times, Singularitarianism relies on mere possibilities: Some form of artificial superintelligence could develop, couldn’t It is awkward and a bit embarrassing to admit, but average it? Yes, it could. But this is a mere logical possibility, that philosophy does not do well with nuances. It may fancy is, to the best of our current and foreseeable knowledge precision and very finely cut distinctions, but what it really there is no contradiction in assuming the development of a loves are polarizations and dichotomies. Internalism or superintelligence. It is the immense between “I , foundationalism or coherentism, trolley left could be sick tomorrow” when I am already not feeling too or right, zombies or not zombies, observer-relative or well, and “I could be a butterfly that dreams to be a observer-independent, possible or impossible worlds, being.” There is no contradiction in assuming that a relative grounded or ungrounded, . . . philosophy may preach the of yours you never heard of just died leaving you $10m. inclusive vel but too often indulges in the exclusive aut aut. Yes, he could. So? Contradictions are never the case, but Such an ability to reduce everything to binary alternatives non-contradictions can still be dismissed as utterly crazy. means that anyone dealing with the continuum of real numbers (pun intended) is likely to be misunderstood. When conditionals and modalities are insufficient, then Singularitarians, often moved, I like to believe, by a sincere The current debate about artificial intelligence (AI) is a sense of apocalyptic urgency, mix faith and facts. They case in point. Here the dichotomy is between believers start talking about job losses, digital systems at risks, and disbelievers in true AI. Yes, the real thing, not Siri in and other real and worrisome issues about computational your iPhone or Roomba in your kitchen. Think instead of the technologies dominating increasing aspects of human false Maria in Metropolis (1927), Hal 9000 in Space Odyssey life, from learning to employment, from entertainment to (1968), C3PO in Star Wars (1977), Rachael in conflicts. From this, they jump to being seriously worried (1982), Data in Star Trek: The Next Generation (1987), Agent about being unable to control their next Honda Civic Smith in (1999), or the disembodied Samantha in because it will have a mind of its own. How true AI and Her (2013). You got the picture. Believers in true AI belong superintelligence will ever evolve autonomously from the to the Church of Singularitarians. For lack of a better term, I skill to park in a tight spot remains unclear, but you have shall refer to the disbelievers as members of the Church of been warned, you never know, and surely you better be AItheists. Let’s have a look at both faiths. safe than sorry.

Singularitarianism is based on three dogmas. First, the Finally, if even this stinking mix of “could,” “if . . . then,” and creation of some form of artificial superintelligence—a “look at the current technologies . . .” does not work, there so-called technological singularity—is likely to happen is the maths. A favourite reference is the so-called Moore’s in the foreseeable future. Both the nature of such a . This is an empirical generalization that suggests that, superintelligence and the exact timeframe of its arrival are in the development of digital computers, the number of left unspecified, although Singularitarians tend to prefer transistors on integrated circuits doubles approximately futures that are conveniently close-enough-to-worry-about every two years. The outcome is more computational power but far-enough-not-to-be-around-to-be-proved-wrong. at increasingly cheaper prices. This has been the case so far, Second, humanity runs a major risk of being dominated by and it may well be the case for the foreseeable future, even such superintelligence. Third, a primary responsibility of the if technical difficulties concerning nanotechnology have current generation is to ensure that the Singularity either started raising some serious manufacturing challenges. does not happen or, if it does, it is benign and will benefit After all, there is a physical limit to how small things can get humanity. As you can see, there are all the elements for a before they simply melt. The problem is that just because Manichean view of the world, with Good fighting against something grows exponentially, this does not mean that , some apocalyptic overtones, the urgency of “we must it develops without boundaries. A great example was do something now or it will be too late,” an eschatological provided by The Economist last November: perspective of human salvation, and an appeal to fears and ignorance. Put all this in a context where people are rightly Throughout recorded history, have worried about the impact of idiotic digital technologies on reigned unchallenged as Earth’s dominant species. their , while the mass media report about new gizmos Might that soon change? Turkeys, heretofore and unprecedented computer disasters on a daily basis, harmless creatures, have been exploding in size, and you have the perfect recipe for a debate of mass swelling from an average 13.2lb (6kg) in 1929 distraction. to over 30lb today. On the rock-solid scientific assumption that present trends will persist, The Like all views based on faith, Singularitarianism is Economist calculates that turkeys will be as big irrefutable. It is also ludicrously implausible. You may more as humans in just 150 years. Within 6,000 years, reasonably be worried about extra-terrestrials conquering turkeys will dwarf the entire planet. Scientists

PAGE 8 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

claim that the rapid growth of turkeys is the result to be wrong. In 1997 he publicly liquefied his article in a of innovations in poultry farming, such as selective food processor and duly drank it. A man of his word. I wish breeding and artificial insemination. The artificial Singularitarians were as bold and coherent as him. nature of their growth, and the fact that most have lost the ability to fly, suggest that not all is lost. Still, I have spent more than a few words to describe with nearly 250m turkeys gobbling and parading Singularitarianism not because it can be taken seriously, in America alone, there is cause for concern. This but because AI disbelievers, the AItheists, can be better Thanksgiving, there is but one prudent course of understood as people over-reacting to all this singularity action: eat them before they eat you.”1 nonsense. I sympathise. Deeply irritated by the worshipping of the wrong digital and the catastrophic prophecies, From Turkzilla to AIzilla, the step is small, if it weren’t for the the Church of AItheism makes its mission to prove once fact that a growth curve can easily be sigmoid (see Figure and for all that any kind of faith in true AI is really wrong, 1), with an initial stage of growth that is approximately totally wrong. AI is just computers, computers are just exponential, followed by saturation, then a slower growth, Turing Machines, Turing Machines are merely syntactic maturity, and finally no further growth. But I suspect that the engines, and syntactic engines cannot think, cannot know, representation of sigmoid curves might be blasphemous and cannot be conscious. End of the story. AI does not for Singularitarianists. and cannot exist. Even bigots should get it. This is why computers (still) cannot do something (the something

being a conveniently movable target), and are unable to process semantics (of any language, Chinese included, no

matter what translation achieves). This proves that

there is absolutely nothing to talk about, let alone worry about. There is no AI, so a fortiori there are no problems

caused by it; relax and enjoy all these wonderful electric

gadgets.

Both Churches seem to have plenty of followers in California,

the place where Hollywood sci-fi films, wonderful research universities like Berkeley, and some of the most important

digital companies in the world live side by side. This may not be accidental, especially when there is a lot of money involved. For example, everybody knows that Google Wiki pe di a , G ra ph of L og i st i c C urve, a ty pi ca l s i gm oid f unct i on Figurehttp://commons.wikimedia.org/wiki/File:Logistic 1. Graph of Logistic Curve,- curve.svg#metadataa typical sigmoid has been buying AI tech companies as if there were no function. Wikipedia, http://commons.wikimedia.org/wiki/ tomorrow (disclaimer: I am a member of Google’s Advisory File:Logistic-curve.svg#metadata Council on the right to be forgotten.4 Surely they must know Enough. I used to think that Singularitarianism was merely funny. Not unlike people wearing tin foil hats. I was wrong, for two reasons. First, plenty of intelligent people have joined the something, with regard to the real chances of developing a Enough.Church: IBill used Gates, Stephen to think Hawking, that or Elon SingularitarianismMusk, Tesla CEO, who has gone as wasfar as to merely computer that can think, that we, outside “The Circle,” are tweet that “We need to be super careful with AI. Potentially more dangerous than nukes.” I funny.guess Not we shall unlike be safe from people true AI as long wearing as we keep using tin Windows foil hats. but, sadly, I suchwas wrong, missing. Thus, Eric Schmidt, Google Executive Chairman, testimonials have managed to transform a joke into a real concern. Second, I have realized for twothat Singularitarianism reasons. is irresponsiblyFirst, plenty distracting. Itof is a richintelligent-world preoccupation, people likely to have speaking at The Aspen Institute on July 16, 2013, stated, joinedworry the people Church:in leisure societies, Bill who Gates,seem to forget Stephen what real are Hawking, oppressing humanity or Elon “Many people in AI believe that we’re close to [a computer and our planet, from environmental disasters to financial crises, from religious intolerance Musk,and violentTesla terrorism CEO, to famine, who poverty, has ignorance, gone and appalling as far living asstandards, to justtweet to that passing the Turing Test] within the next five years.”5 I do not mention a few. Oh, and just in case you thought predictions by experts were a reliable guide, “We thinkneed twice . Thereto arebe many super staggeringly careful wrong technological with predictions AI. Potentially by great experts more know who the “many” are, but I know that the last people dangerous(see some hilarious than ones nukes.” in (Pogue 18 JanuaryI guess 2012) andwe (Cracked shall Readers be 27 safe January from true you should ask about whether something is possible are 2014)). For example, in 2004 Bill Gates stated “Two years from now, spam will be solved.” AI asAnd long in 2011 Stephenas we Hawking keep declared usingthat “philosophy Windows is dead” (Warman but, 17 May sadly, 2011), such those who have abundant financial reasons to reassure you so you are not reading this article. But the prediction of which I am rather fond is by Robert testimonialsMetcalfe, co-inventor have of Ethernet managed and founder to of 3Com transform. In 1995 he promised a joke to “eat hisinto a real that it is. So let me offer a bet. I hate aubergine (eggplant), concern.words” ifSecond, his prediction that I “thehave Internet realized will soon go supernova that andSingularitarianism in 1996 will is but I shall eat a plate full of it if a software program will get catastrophically collapse” should turn out to be wrong. In 1997 he publicly liquefied his irresponsiblyarticle in a food processordistracting. and duly drank It it. Ais man a of rich-world his word. I wish Singularitarians preoccupation, were the gold medal (i.e., pass the Turing Test) of a Loebner Prize as bold and coherent as him. likely to worry people in leisure societies, who seem to competition before July 16, 2018. It is a safe bet. So far, we forgetI have what spent more real than a fewevils words toare describ oppressinge Singularitarianism no humanityt because it can be takenand our have seen only consolation prizes given to the less badly seriously, but because AI disbelievers, the AItheists, can be better understood as people over- planet,reacting from to all this environmentalsingularity nonsense. I sympathise. disasters Deeply irritated to by thefinancial worshipping of crises, performing versions of contemporary ELIZA. As I explained the wrong digital gods and the catastrophic prophecies, the Church of AItheism makes its frommission religious to prove once intolerance and for all that any kindand of faith violent in true AI isterrorism really wrong, totally to famine, when I was a judge the first time the competition came to poverty,wrong. AIignorance, is just computers, computersand appalling are just Turing Machines, living Turing standards, Machines are just to the UK, it is human interrogators who often fail the test, by mention a few. Oh, and just in case you thought predictions asking binary questions such as “Do you like ice cream?” by experts were a reliable guide, think twice. There are or “Do you believe in ?” to which any answer would be many staggeringly wrong technological predictions by utterly uninformative in any case.6 I wonder whether Gates, great experts.2 For example, in 2004 Bill Gates stated “Two Hawking, Musk, or Schmidt would like to accept the bet, years from now, spam will be solved.” And in 2011 Stephen choosing a food of their dislike. Hawking declared that “philosophy is dead,” so you are not reading this article.3 But the prediction of which I am rather Let me be serious again. Both Singularitarians and AItheists fond is by Robert Metcalfe, co-inventor of Ethernet and are mistaken. As Alan Turing clearly stated in the article founder of 3Com. In 1995 he promised to “eat his words” where he introduced his famous test (Turing 1950), the if his prediction that “the Internet will soon go supernova question “Can a machine think?” is “too meaningless to and in 1996 will catastrophically collapse” should turn out deserve discussion” (ironically, or perhaps presciently, that

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 9 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

question is engraved on the Loebner Prize medal). This It may all seem rather commonsensical. But if you try to holds true, no matter which of the two Churches you belong explain it to an AItheist like John Searle he will crucify you to. Yet both Churches dominate this pointless debate, together with all the other Singularitarians. In a review of suffocating any dissenting voice of . True AI is not my book, The Fourth Revolution – How the Infosphere is logically impossible but it is utterly implausible. According Reshaping Humanity, where I presented some of the ideas to the best of our scientific knowledge today, we have no above, Searle criticized me for being a believer in true AI idea how we may begin to engineer it, not least because and a metaphysician who thinks that reality is intrinsically we have very little understanding of how our brain and our informational.8 This is nonsense. As you might have guessed own intelligence work. This means that any concern about by now, I subscribe to neither thesis.9 In fact, there is much the appearance of some superintelligence is laughable. I agree about with Searle’s AItheism. So I tried to clarify my What really matters is that the increasing presence of ever- position in a reply.10 Unsuccessfully. Unfortunately, when smarter technologies in our lives is having huge effects on people react to Singularitarianism, to blind faith in the how we conceive ourselves, the world, and our interactions development of true AI or to other technological fables, they among ourselves and with the world. The point is not that run the risk of falling into the opposite trap and thinking that our machines are conscious, or intelligent, or able to know the debate is about computers (it is not—social media and something as we do. They are not. The point is that they are , for example, are two major issues in the philosophy increasingly able to deal with more and more tasks better of information) and that these are nothing more than than we do, including predicting our behaviors. So we are not electric typewriters, not worth a philosophical investigation. the only smart agents around, far from it. This is what I have They swing from the pro-AI to the anti-AI, without being defined as the fourth revolution in our -understanding. able to stop, think, and reach the correct, middle ground We are not at the center of the universe (Copernicus), of the position, which identifies in the information revolution a biological kingdom (Darwin), or of the realm of rationality major transformation in our Weltanschauung. Let me give (Freud). After Turing, we are no longer at the center of the you some elementary examples. Our self-understanding world of information and smart either. We share has been hugely influenced by issues concerning privacy, the infosphere with digital technologies. These are not the right to be forgotten, and the construction of personal the children of some sci-fi superintelligence, but ordinary identities online. Just think of our idea of friendship in artefacts that outperform us in ever more tasks, despite being a world dominated by social media. Our interactions no cleverer than a toaster. Their abilities are humbling and have hugely changed due to online communications. make us revaluate our intelligence, which remains unique. Globalization would be impossible without the information We thought we were smart because we could play chess. revolution, and so would have been many political Now a phone plays better than a chess master. We thought movements, or hacktivism. The territoriality of the law has we were free because we could buy whatever we wished. been completely disrupted by the onlife (sic) world, in Now our spending patterns are predicted, sometimes even which online and offline experiences are easily continuous, anticipated by devices as thick as a plank. What does all this thus further challenging the Westphalian system.11 Today mean for our self-understanding? science is based on Big Data and , simulations and scientific networks, all aspects of an that The success of our technologies largely depends on the is massively dependent on, and influenced by, information fact that, while we were speculating about the possibility technologies. Conflicts, crime, and security have all been of true AI, we increasingly enveloped the world in so re-defined by the digital, and so has political power. In many devices, applications, and data that it became short, no aspect of our lives has remained untouched by an IT-friendly environment, where technologies can the information revolution. As a result, we are undergoing replace us without having any understanding or semantic major philosophical transformations in our views about skills. Memory (as in algorithms and immense datasets) reality, ourselves, our interactions with reality, and among outperforms intelligence when landing an aircraft, finding ourselves. The information revolution has renewed old the fastest route from home to the office, or discovering philosophical problems and posed new, pressing ones. the best price for your next fridge. The BBC has made a This is what my book is about, yet this is what Searle’s two-minutes short animation to introduce the idea of a review entirely failed to grasp. fourth revolution that is worth watching.7 Unfortunately, like John Searle, it made a mistake in the end, equating I suspect Singularitarians and AItheists will continue their “better at accomplishing tasks” with “better at thinking.” diatribes about the possibility or impossibility of true AI for I never argued that digital technologies think better than the time being. We need to be tolerant. But we do not have us, but that they can do more and more things better than to engage. As Virgil suggests to Dante in Inferno, Canto III: us by processing increasing amounts of data. What’s the “don’t mind them, but look and pass.” For the world needs difference? The same as between you and the dishwasher some good philosophy and we need to take care of serious when washing the dishes. What’s the consequence? That and pressing problems. any apocalyptic vision of AI is just silly. The serious risk is not the appearance of some superintelligence, but that we NOTES may misuse our digital technologies, to the detriment of a 1. “Turkzilla!” The Economist. large percentage of humanity and the whole planet. We are 2. See some hilarious ones in Pogue, “Use It Better,” and and shall remain for the foreseeable future the problem, Cracked Readers. not our technology. We should be worried about real human 3. Matt Warman, “ Tells Google ‘Philosophy Is stupidity, not imaginary artificial intelligence. The problem Dead.” is not HAL but H.A.L., Humanity At Large.

PAGE 10 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

4. Robert Herritt, “Google’s Philosopher.” 5. https://www.youtube.com/watch?v=3Ox4EMFMy48 First-Person Consciousness as Hardware 6. Luciano Floridi, Mariarosario Taddeo, and Matteo Turilli, “Turing’s Imitation Game.” Peter Boltuc UNIVERSITY OF ILLINOIS SPRINGFIELD AND AUSTRALIAN 7. http://www.bbc.co.uk/programmes/p02hvcjm NATIONAL UNIVERSITY 8. John R. Searle, “What Your Computer Can’t Know.” 9. The reader interested in a short presentation of what I mean by INTRODUCTION informational realism may wish to consult Floridi, “Informational I take the paradigmatic case of first-person consciousness to Realism.” For a full articulation and defense, see Floridi, The Philosophy of Information. be when a nurse says that a patient regained consciousness after surgery. The patient does not need to have memory 10. Floridi, “Response to NYROB Review.” or other advanced cognitive functions. But she is online, so 11. Floridi, The Onlife Manifesto. to say—we have good reasons to believe that the question what it is like for her to be is not empty. BIBLIOGRAPHY Cracked Readers. “26 Hilariously Inaccurate Predictions about Advanced cognitive architectures, such as LIDA, approach the Future,” January 27, 2014. http://www.cracked.com/ the functional threshold of consciousness. Such software photoplasty_777_26-hilariously-inaccurate-predictions-about-future/. performs advanced cognitive functions of all kinds, Floridi, Luciano. “Response to NYROB Review.” The New York Review of Books, November 20, 2014. http://www.nybooks.com/articles/ including image making and manipulation, advanced archives/2014/dec/18/information-desk/. memory organization and retrieval, communication Floridi, Luciano. 2003. “Informational Realism.” Selected papers from (including semantic structures), perception (that includes conference on Computers and Philosophy, volume 37. responses to color, temperature, and other ), and Floridi, Luciano. The Philosophy of Information. Oxford: Oxford even creativity (e.g., imagitrons). Some AI experts believe University Press, 2011. that, at a certain threshold, adding further cognitive Floridi, Luciano. The Fourth Revolution: How the Infosphere Is Reshaping functions would result in first-person consciousness. Non­ Human Reality. Oxford: Oxford University Press, 2014a. reductivists claim that the latter would emerge based on Floridi, Luciano, ed. The Onlife Manifesto: Being Human in a an informationally rich emergence base. Reductivists claim Hyperconnected Era. New York: Springer, 2014b. that such a rich information processing structure just is Floridi, Luciano, Mariarosaria Taddeo, and Matteo Turilli. “Turing’s consciousness, that there is no further fact of any kind. I Imitation Game: Still a Challenge for Any Machine and Some Judges.” disagree with both claims. and Machines 19, no. 1 (2009): 145–50. Herritt, Robert. “Google’s Philosopher.” Pacific Standard, December The kind of first-person consciousness in the example of a 30, 2014. http://www.psmag.com/nature-and-technology/googles­ philosopher-technology-nature--court-legal-policy-95456. patient regaining consciousness is analogous to a stream of light—it is not information processing of some advanced Pogue, David. “Use It Better: The Worst Tech Predictions of All Time – Plus, Flawed Forecasts about Apple’s Certain Demise and the Poor sort. Just like light bulbs are pieces of hardware, so are the Prognostication Skills of Bill Gates,” January 18, 2012. http://www. parts of animal brain that create first-person consciousness.1 scientificamerican.com/article/pogue-all-time-worst-tech-predictions/. Every object can be described as information (Floridi) and Searle, John R. “What Your Computer Can’t Know.” The New York is in principle programmable (a physical interpretation of Review of Books, October 9, 2014. http://www.nybooks.com/articles/ Church-Turing thesis), but this does not make every object archives/2014/oct/09/what-your-computer--know/. in the universe a piece of software. The thesis of this paper The Economist. “Turkzilla!” November 27, 2014. http://www.economist. is that first-person consciousness is more analogous to a com/blogs/graphicdetail/2014/11/daily-chart-16. piece of hardware, a light emitting bulb, than to software. Turing, A. M. “Computing Machinery and Intelligence.” Mind 59, no. There are probably information processing thresholds 236 (1950): 433–60. below which first-person consciousness cannot function Warman, Matt. “Stephen Hawking Tells Google ‘Philosophy Is Dead’.” (just like a bulb cannot emit light if not hooked up to The Telegraph, May 17, 2011. http://www.telegraph.co.uk/technology/ google/8520033/Stephen-Hawking-tells-Google-philosophy-is-dead. the source of electricity), but no amount of information html. processing, no cognitive function, shall produce first- person consciousness without such consciousness emitting a piece of hardware.

This claim follows from the so-called thesis, the idea that if first-person consciousness is a natural process it needs to be replicable in . Instituting such functionality in machines would require a special piece of hardware, slightly analogous to the projector of holograms. On the other hand, human cognitive functions can be executed in a number of cognitive architectures.2 Such architectures do not need to be hooked up to the light­ bulb-style first-person consciousness. This last claim opens the issue of philosophical zombies and . On both issues I try to keep the course between Scylla and Charybdis presented by the most common alternatives.

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 11 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

THE ENGINEERING THESIS irreducible first-person consciousness, do not have this In recent works I advanced the engineering thesis in stream to project their phenomenal information onto. The machine consciousness (Boltuc 2012, 2010, 2009; Boltuc sub-system of CNS responsible for producing the stream of and Boltuc 2007).3 The argument goes as follows: proto-consciousness (Damasio), is a piece of hardware, just like a light bulb belongs to hardware.6 Also, the light, which 1) Assume that we accept the non-reductive theory of is a stream of photons, is much like hardware, similar to the consciousness. stream of water, though some ontologists may disagree due to the peculiar (wave-corpuscular) status of light.7 2) Assume that we are physicalists (non-reductive materialists broadly defined). Due to the fact that first-person consciousness is not just information processing, it should be viewed as hardware. => Information (a product of software) gets embroiled in the stream of first-person consciousness as the conscious 3) First-person consciousness must be generated by being becomes more and more conscious of things through some natural mechanism, probably in animal brains. information processing.

If one accepts some version of —instead It is not clear whether the conscious element helps of “produced”—consciousness is collected or enhanced information processing in any way, though it is plausible by brains. that it does (just like light helps viewers see details in the room). Below we explore whether first-person -> From 3 and historic regularity of development of science consciousness is merely epiphenomenal—in some detail.

4) One day, as neuroscience develops, we should get to EPIPHENOMENALISM REVISITED know how first-person consciousness works. Is first-person consciousness just information processing? If it is, then its operation can be described by an . 5) To know well how anything is produced in nature is Such algorithms could be followed by non-conscious to understand in detail how such producing occurs. AI engines. (To be sure, such AIs would be functionally To have such an understanding means to have an conscious. Yet, they would not be first-person conscious in engineering blueprint of the process. terms of non-reductive consciousness.) The question arises: Is first-person consciousness merely epiphenomenal? 6) Once we have an engineering blueprint of first-person consciousness, we should, in principle, be able to There are two ways to address this question. generate it. A) To claim that non-reductive consciousness does => something that purely functional consciousness could not do. If so, consciousness would not be 7) We should be able to engineer first-person epiphenomenal. I discuss the light version of this claim: consciousness. Consciousness and, in particular, qualia, bring about a way to mark certain states of affairs, which happen This argument helps us avoid anthropocentric , to be optimal in cognitive architectures of advanced the claim that first-person consciousness is physical but animals. in some important sense reserved for human and select animals. If first-person consciousness is natural it B) To bite the bullet and accept that first-person must, in principle, be implementable in artificial objects.4 consciousness does nothing in functional terms. If so, consciousness would be epiphenomenal. I discuss, CONSCIOUSNESS AS HARDWARE and provisionally endorse, the indirectly relevant It should now be clear that Turing was right; there are version of this claim: While first-person consciousness no functionalities that AI is unable to replicate (at the does not perform any unique functions, we have right level of granularity). Functional consciousness is reasons to care whether certain organisms have or lack the programming that allows one to perform cognitive such consciousness. Those reasons are moral reasons, functions. It is rightly viewed as software. First-person in a broad sense of the term. consciousness also tends to be viewed as software that processes specific phenomenal information, but it should A) THE NON-EPIPHENOMENAL ALTERNATIVE: 5 not. QUALIA AS MARKERS Phenomenal information, just like any information, can I used to argue that first-person consciousness should be be processed by robots with no irreducible first-person viewed as a convenient marker, maybe even a unique one consciousness. First-person consciousness should (more likely non-unique but best available).8 By a marker I rather be viewed as analogous to a stream of light or mean something like color-coding. Your can code files on a holographic projection, though those analogies are your desktop by different symbols or shades of gray, but remote. Some functionally conscious entities have it, and the color coding makes the coding easily recognizable to their information processing is first-person conscious. the human eye, the eyes of many animals, and some of Other functionally conscious entities, those with no the non-animal preceptors. Phenomenal consciousness

PAGE 12 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

allows us to use colors, scents, sounds, and other qualia zombies opens an interesting problem: How rough can a in a way that is at least as good, and for human cognitive zombie get?12 architecture better than, the other potential kinds of coding (say, using the electron spin). This argument was my last Let me explain. Chalmers argues that it is conceivable that, ditch effort to do two things: save qualia as essential to for two physically identical , one is a zombie first-person consciousness, and also view them as a way to while the other has first-person consciousness. Dennett secure its non-epiphenomenal status. responds that such an assumption violates the very tenet of (there is no difference without physical Gradually, I have been losing faith in this two-step effort. difference) and therefore begs the question, if the zombie I still retain some for this approach but I doubt argument is to be used in polemics against . that it works. The main reason in favor of the approach is an I think Dennett is right since the argument begs the analogy with light (a different analogy than the one used question.13 An interesting task is to define the zombie most elsewhere in this paper)—the light reflected or absorbed similar to a first-person conscious human being that does by objects enables us to gain visual information; it is not violate the claim that there is no difference without not identical with such information but it is, usually, its physical difference. To use David Lewis’s ontology of necessary condition. possible worlds, the goal is to establish the closest in which zombies dwell. Well, if functionally—in The main reason against this approach is the following: After terms of information processing—zombies and first-person some conversations with , contrary to his conscious individuals would have the same cognitive intentions, I lost faith in the idea that the hard-problem of abilities, the only difference would be that the latter have consciousness is the problem of experience. To be precise: a certain “projector of consciousness.” Such a projector If Chalmers’s hard-problem is the problem of experience, would have to have a physical basis. Probably the smallest then my problem of first-person consciousness is not the possible difference could be attained if both the zombies hard problem since it is not the problem of experience. Why and the non-zombies would have a (physical) projector of not? If we carefully read a standard paper on phenomenal consciousness—functionally analogous to the projector of consciousness for robots (say, Franklin et al., “A holograms, or to the projector of light (one such projector Phenomenally Conscious Robot?”) we can see that there is is a light bulb). In terms of the zombies such a projector a notion of purely functional reaction of robots, or humans, would not function, and the malfunction would be caused to sound, color, smell, and other phenomenal qualia. The by the smaller possible error—by something like a burn-out robots have functional-phenomenal consciousness. What of a small wire that prevents the functioning of a light bulb. distinguishes their phenomenal consciousness from the other kind of phenomenal experience, namely, from the Here is a way to present the argument of this paper based first-person consciousness, is that those who possess the on the issue at hand: The light bulbs and projectors of latter have the first-person subjective feel of qualia. Their holograms are pieces of hardware, and so are the brain­ information processing of phenomenal information seems cells most likely responsible for generation of first-person exactly the same, or at least very similar. This conclusion consciousness. The first avenue to take—to maintain that can also be drawn from the physical interpretation of first-person consciousness affects information processing— the Church-Turing thesis. Hence, there are two kinds of has something to its advantage, but the above discussion phenomenal experience and only one of them relates to of zombies leads to the second approach: the approach the hard problem of consciousness. Block seems to make that first-person consciousness is epiphenomenal. a similar distinction, though not very prominently. C) THE EPIPHENOMENAL ALTERNATIVE: To conclude: The informational structure of phenomenal FIRST-PERSON CONSCIOUSNESS IS INDIRECTLY qualia is NOT what makes a difference between reductive and non-reductive approaches. The difference is in the RELEVANT irreducible first-person perspective on phenomenal The second approach to non-reductive consciousness information that humans have and AI engines lack, at least endorses epiphenomenalism. Most philosophers would these days. scoff at the idea; epiphenomenalism seems hardly worth any respect. If first-person consciousness does not do B) A ZOMBIE INTERLUDE anything, it is practically irrelevant and empirically not­ The above conclusion makes qualia-based arguments verifiable—two bummers, or so it seems. Yet, there is at irrelevant (or, rather, not directly relevant) to the hard least one aspect such that first-person consciousness is problem of consciousness. For instance, Jackson’s Black relevant even if it is functionally epiphenomenal. and White Mary argument tells us something important about human cognitive architecture;9 it tells us that we The epiphenomenal does not need to mean irrelevant. have no connection from knowledge by description to the Imagine a sex robot that behaves just like a human lover, actual sensors of colors and other qualia in the brain.10 The at the relevant level of granularity, but has no first-person argument—so reformulated—is not directly relevant for the consciousness. I think it should matter whether one’s debate of irreducible first-person consciousness since it lover, or a close friend, merely behaves as if he/she had relates to specificity of human cognitive architecture. So first-person consciousness or whether he/she in fact has does the .11 The case of zombies is relevant first-person consciousness.In response to this point, for the argument advanced in this paper for the reasons Alan Hajek pointed out that whether one’s friend has first- that may not be the gist of the zombie case. The issue of person consciousness should matter even more outside of

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 13 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

the context of sex. This is a persuasive point but maybe In his recent work the author moves a step further, and less so for those philosophers who do not endorse first- maybe a little too far.15 He starts questioning the theory of person consciousness already. For them this general evolution, not by pointing out that maybe it requires some question may be viewed as meaningless or speculative fixes but by posing that we may need to reject the gist of it (for instance, due to the problem of privileged access). The and engage in some teleological theory of a mind, or spirit, cultural expectations that one should care whether one’s with the purpose creating the world.16 Nagel expresses his lover actually feels his/her love or just behaves as if she amazement in human cognitive powers, and consciousness, or he did seem to play a role in this context, and they may and claims that they would not have emerged from chance be stronger than the other epistemic intuitions. This is in and randomness. All this is happening today, when science fact a bit strange. It may come, in part, from the fact that provides quite good hypotheses of how consciousness people in love are supposed to connect with one another evolved (Damasio). He also seems to disregard the older in a manner not prone to verificationist objections; another sound approaches showing how order and life emerge explanation may come from the fact that psychology of from chaos (Monod). Nagel’s disappointing change in view most epistemic functions used by reductionists harkens puts into question the gist of non-reductive naturalism. back on mid-twentieth-century philosophy of science (Popper) whereas psychology of sex and love follows a Also, David Chalmers abandoned non-reductive materialism. different, more intuitively plausible paradigm. In the past Chalmers presented a number of potential theories in and desisted from making a If I care about whether my child, my friend, or my lover is choice among them (Chalmers). He kept open the possibility in fact feeling the world, or my interaction with her or him, of non-reductive materialism as well as panpsychism. I I have a legitimate interest in whether an individual does viewed this work as an example of intellectual honesty and or does not have first-person consciousness despite one’s the ability to overcome human psychological tendencies exact same external functioning. Hence, I have shown at to drive towards hasty conclusions. A few years back, least one broad of instances when epiphenomenalism Chalmers endorsed panpsychism, moreover in its dualistic about first-person consciousness does not lead to an form. He accepted the idea that the mental substance is irrelevant question. The question is even more relevant if one of the elements in the world potentially available to we have a way of discovering strong inductive evidence science, but that it is essentially different from the material. whether one has or lacks first-person consciousness. Such This dualistic approach differs from neutral as evidence would be missing in the world of zombies. In the another form of panpsychism (formulated by Spinoza); not world of rough zombies, as we have seen above, while (at to mention basically materialistic neutral monism presented a given level of granularity) there may be no difference in by Russell (1921). functioning between a zombie and a being with first-person consciousness, there is a physical difference between the What are the background reasons for those radical choices two: the non-zombie has a unit (projector of consciousness) of at least two of the former top champions of non- that, if properly functioning, does produce consciousness, reductive physicalism or materialism? If we were to look whereas zombies do not have such a functioning unit. for the common denominator of Nagel’s and Chalmers’s Hence, first-person consciousness matters even if it does decisions, it is their robust, inflationary, idea of the not influence any functionalities. Moreover—as we see subject of consciousness. Many philosophers tend to both from the rough zombies argument and from the view certain aspects of personal being as essential parts engineering thesis—it can be empirically verifiable (by of the subject, or consciousness. However, thinking, even inductive methods) which individuals have and which ones creative thinking, memory, color and smell recognition, or lack the capacity for producing consciousness, and in fact emotional states (in their functional aspect) are features whether such capacity is activated—this translates into of human cognitive architecture that are programmable them having first-person consciousness. in a robot, or some other kind of a zombie. They are, by themselves, just software products. DEFLATIONARY MOTIVATION There is another reason to adopt a very weak theory of non- If we want to find something unique, as non-reductive reductive consciousness. A deflationary approach may be philosophers should, we ought to dig more deeply. All the best, or only, chance to save non-reductive physicalism. information processing, whether it is qualia perception, thinking and memory, or creative processes, can be once made a very important point: It is a programmed and therefore is a part of the content—of better heuristic hypothesis to assume that we know 20 an object defined as content, as some functionalities. percent of what there is to know than the 80 or 90 percent By physical interpretation of the Church-Turing thesis, that many scientists and philosophers tend to assume.14 such content can always be represented in mathematical There is no reason to assume that if human civilization lasts functions that, almost certainly, can be instantiated by another few thousand years we will stop making crucial other means in other entities. The true subjectivity is not discoveries in basic sciences. Those discoveries, if they software at all; it is the stream of awareness before it are as big as Einstein’s revolution, add up to a justification even reflects any objects we are aware of. Let us come of the new ways of thinking that may be inconsistent with back to the story of a patient in a hospital when a nurse some important aspects of what we consider a scientific discovers that he or she regained consciousness, even view today. All of this did not prevent Nagel from claiming though we may be unsure of what he or she is aware of. to endorse non-reductive materialism. Until recently, that Such consciousness, just like a stream of water, or some is. Roentgen rays, or any other sort of light—is not a piece

PAGE 14 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

of software. It is hardware. That internal light, to use an 11. Details in the upcoming book Non-reductive Consciousness: old-fashioned sounding phrase, is the gist—and, in fact, Naturalistic Deflationary Approach. it is the whole shebang—of what is non-reductive in non- 12. This is the title of an existing paper I presented at various venues reductive naturalism. Any and all information processing in 2014. can be duplicated in cognitive architectures with no first- 13. I leave aside Chalmers’s intricate argument that proceeds from person non-reductive consciousness (in zombies, if one conceivability to modally stronger notions. I think Chalmers is successful in showing that there is a plausible modal language likes this theatrical term). (system of modal logic) in which zombies can be defended. I also think Dennett shows that such language may not be used in This is my controversial claim: First-person consciousness debate with reductive physicalism. is not like a piece of software but of hardware. This view 14. Nagel, Mortal Questions; Nagel, The View from Nowhere. may look like a version of type E dualism since such dualism 15. Nagel, Mind and Cosmos. is linked to epiphenomenalism about consciousness. Yet, it 16. I think this is what may be called the Spencer trap. In his attempt would be difficult to interpret as dualism a position that to endorse evolutionary theory and implement it to all matters, consciousness is as material as hardware. (A view that Spencer made scientific claims from a philosophical standpoint. maintains that software is material but hardware is not Nagel seems to follow a similar methodology to the opposite would be really quite odd, wouldn’t it?) effect. BIBLIOGRAPHY TO SUM UP I began with an argument that first-person consciousness Block, N. “On a Confusion about a Function of Consciousness.” Behavioral and Brain Sciences 18, no. 2 (1995): 227–87. should be a natural process and that we should be able to ———. “Consciousness.” In Oxford Companion to the Mind, 2nd ed., engineer it in machines (the engineering thesis). But first- edited by R. Gregory. Oxford University Press, 2004. person consciousness is not just an information-processing Boltuc, P. “The Engineering Thesis in Machine Consciousness.” Techné: mechanism. First-person consciousness lies beyond any Research in Philosophy and Technology 16, no. 2 (Spring 2012): 187– information processing. The fact that it is not information 207. processing and not a functionality of any sort makes ———. “What Is the Difference between Your Friend and a Church the first-person consciousness unique and irreducible. Turing Lover?” In The Computational Turn: Past, Presents and Futures?, Thanks to the recent works in cognitive neuroscience and 37–40. C. Ess; R. Hagengruber Aarchus University, 2011. psychology, the view of non-reductive consciousness as ———. “A Philosopher’s Take on Machine Consciousness.” In hardware seem better grounded than the alternatives. Philosophy of Engineering and the Artifact in the Digital Age, edited by V. E. Guliciuc, 49–66. Cambridge Scholar’s Press, 2010. ACKNOWLEDGEMENTS ———. “The Philosophical Problem in Machine Consciousness.” International Journal of Machine Consciousness 1 (2009): 155–76. I am grateful to Rachel Briggs and David Chalmers for good discussions and encouragement. ———. “Mary’s Acquaintance.” APA Newsletter on Philosophy and Computers 14, no. 1 (2014): 25–31. NOTES Boltuc P., and N. Boltuc. “Replication of the Hard Problem of Consciousness in AI and Bio-AI: An Early Conceptual Framework.” In AI 1. Whether light is hardware is an interesting topic in ontology, but and Consciousness: Theoretical Foundations and Current Approaches, it is definitely not software. A. Chella, R. Manzotti, 24–29. Merlo Park, CA: AAAI Press, 2007. Also 2. I actually think all human cognitive functions, though this is online: http://www.Consciousness.it/CAI/online_papers/Boltuc.pdf. a stronger claim than I may need for the sake of the current Chalmers, D. Facing Up to the Problem of Consciousness Journal of argument. Consciousness Studies 2, no. 3 (1995): 200–19. 3. Boltuc, “The Engineering Thesis in Machine Consciousness”; Damasio, A. Self Comes to Mind. Constructing the Conscious Brain. Boltuc, “A Philosopher’s Take on Machine Consciousness”; Boltuc, 2010. “The Philosophical Problem in Machine Consciousness”; Boltuc and Boltuc, “Replication of the Hard Problem of Consciousness Dennett, D. . Boston: The Penguin Press, 1991. in AI and Bio-AI.” ———. “The Unimagined Preposterousness of Zombies.” Journal of 4. It is an open question whether it requires carbon-based, organic, Consciousness Studies 2, no. 4 (1995a): 322–26. chemistry. Franklin, S., B. Baars, and U. Ramamurthy. “A Phenomenally Conscious Robot?” APA Newsletter on Philosophy and Computers 8, no. 1 (Fall 5. This is the standard AI approach. See Franklin, but also the works 2008): 2–4. Available at http://www.apaonline.org/publications/ by Aaron Sloman, Igor Alexander, and others. newsletters/v08n1_Computers_03.aspx. 6. Proto-consciousness is not identical to stream of consciousness; Monod, J. Chance and Necessity. New York: Alfred A. Knopf, 1981. it is more of a stable background for cognitive tasks, but the task of drawing an exact analogy with neuroscience is one for another Nagel, T. Mind and Cosmos. Why the Materialist Neo-Darwinian article. Conception of Nature is Almost Certainly False. Oxford University Press, 2012. 7. Still, they would disagree even more strongly with the claim that light is just a piece of software. ———. The View from Nowhere. Oxford University Press, 1986. 8. Boltuc, “A Philosopher’s Take on Machine Consciousness.” ———. Mortal Questions. Oxford University Press, 1979. 9. Boltuc, “Mary’s Acquaintance.” Russell, B. The Analysis of Mind. : George Allen and Unwin; New York: The Macmillan Company, 1921. 10. The link goes one way, from experience to description. One could bio-engineer the reverse link, but evolution left us without it since knowledge by description is evolutionarily new.

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 15 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

the topic, that “Nobody puts their membership in Alcoholics Social Media and the Organization Man Anonymous on their CV.”3 Surely we ought to say that if anything demonstrates integrity, it would be admitting D. E. Wittkower a difficult truth about oneself and seeking support with OLD DOMINION UNIVERSITY others through a frank relationship of self-disclosure, making the AA example particularly apt, not least since the In an age of social media we are confronted with a problem “anonymous” part of AA recognizes that this sort of integrity novel in degree if not in kind: being called to account for requires a safe separation of this organizational identity the differences between presentations of self appropriate from other aspects of one’s life, of which the contents of a within a variety of group contexts. Business news in the CV is only one particular example, dramatic in its absurdity. post- era has been replete with stories about privacy fails, large and small—employees fired or denied Zuckerberg, for his part, seems to have started to think promotion seemingly due to same-sex relationships differently about this, stating in a 2014 interview that revealed on social media, career advice to college students about destroying online evidence of having done normal I don’t know if the balance has swung too far, but college-student things, and so on. Keeping work and I definitely think we’re at the point where we don’t private lives separate has become more difficult, and need to keep on only doing real identity things difficult in different ways, and we are living in a new era of [. . .] If you’re always under the pressure of real navigating self- and group-identities. identity, I think that is somewhat of a burden.4

While social media in general tends to create these The 2010 comments are still important for us to take problems, Facebook, with its unitary profile, single Friend seriously, though. Not so much because Zuckerberg’s list, and real-name policy, has been central to creating this comments reveal a design trait in the Facebook platform new hazardous environment for identity performance. Mark that has changed how we think about and perform identity Zuckerberg is quoted in an interview with David Kirkpatrick (although this is interesting as well!). But even more so saying, “You have one identity. . . . The days of you having because if Zuckerberg, mired as he is in thinking about a different image for your work friends or co-workers and how people manage self- and group identities, can fall into for the other people you know are probably coming to an a way of thinking so disconnected from the actual conduct end pretty quickly. . . . Having two identities for yourself is of lives, there must be something deeply intuitive, perhaps an example of a lack of integrity.”1 Many have critiqued this seductive, about this way of thinking about integrity. simplistic view of identity, but Michael Zimmer’s widely read blog post on the topic is particularly pithy and direct: At the heart of this is a modern individualist notion of the self—the self which -bearing, with an Zuckerberg must have skipped that class individual and separable existence; the juridical self. We where Jung and Goffman were discussed. must assume an integral self logically prior to organizational Individuals are constantly managing and restricting and communal entanglement in order to pass judgment flows of information based on the context they on whether it is limited, transformed, disfigured, hidden, are in, switching between identities and persona. or altered by its entrance into and representation within I present myself differently when I’m lecturing in groups and contexts. We tend to take on a “correspondence the classroom compared to when I’m having a beer theory” of integrity, parallel to the correspondence theory with friends. I might present a slightly different of truth, in which a self-representation is to have greater identity when I’m at a church meeting compared or lesser integrity depending upon the degree of similarity to when I’m at a football game. This is how we that it bears to some a priori “true” self. This view of an navigate the multiple and increasingly complex “unencumbered self” is deeply mistaken as Sandel (1984) spheres of our lives. It is not that you pretend among others has pointed out, but is logistically central to to be someone that you are not; rather, you turn our liberal individualist conception of rights and community the volume up on some aspects of your identity, and thus hard to avoid falling into. Zuckerberg may do well and tone down others, all based on the particular to read philosophy in addition to the remedial Goffman context you find yourself.2 (1959) to which Zimmer rightly wishes to assign him.

And this view of the complexity of managing self- INTEGRITY AND SELF-PERFORMANCE presentations within different organizational contexts, Turning to philosophical theories of seems destructive as it already is to Zuckerberg’s—well, it’s hard at first unhelpful. Whether, for example, we adopt a body- to say . . . simplistic? Naïve? Unrealistic? Hetero- and Cis­ continuity or mind-continuity theory of identity has only the privileged? Judgmental? All of these, I suppose—at any slightest relevance to what might count as “integrity”—in rate, to Zuckerberg’s faulty view of multiple identities as fact, it seems any perspective on philosophical personal “a lack of integrity,” this view doesn’t even yet consider identity must view “integrity” as either non-optional or that different elements of identity may need to be not impossible; more a metaphysical state than a moral value. merely emphasized or toned down in different contexts, But even within, e.g., the Humean view that the self is no but that integral aspects of identity may need to be hidden more than a theater stage on which impressions appear entirely in some contexts and revealed only in others. in succession5 fails to preclude that there may be some Zimmer is aware of this too, and quotes an appropriately integral self—Hume’s claim applies only to the self as pseudonymous comment on Kieran Healy’s blog post on revealed by , as Kant pointed out in arguing

PAGE 16 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

for the of the transcendental unity of apperception the self that we recognize as ourselves, and inhibitions may (1998); a grammatical necessity, as it were, corresponding in themselves be the product of choice and work. unknowable ways to the noumenal reality, which, however, is not necessarily less real for its unknowability. Indeed, INTEGRITY IN AN ORGANIZATIONAL CONTEXT when we look to Hume’s (2012) theory of moral , we We need not fall into a correspondence theory of integrity or see it is based upon sentiment and sympathy rather than adopt a liberal individualist conception of the self in order following moral rules or calculation, implying that we have to recognize that organizational contexts present problems these acquired and habitual attributes which constitute our for personal integrity. Two primary sorts come immediately moral , even if they are not the “I” of the “I think” to mind: (1) that organizational contexts may exert which accompanies all representations. Even reductive and influences rendering it more difficult to act with integrity, as skeptical positions within philosophical theories of personal in familiar cases such as conformity and groupthink; and (2) identity make room for habit, character, and some sort of that organizational contexts may contain hostility towards content to the self, inaccessible through introspection certain self-identifications, making self-performance with though it might be, which is subject to change and growth, integrity dangerous. The second kind of problem is the sort and which is if not an origin then at least a conditioning most obviously presented by social media in novel ways, factor in the determination of our thought and action. and will be our focus here, but by the end of this chapter we’ll have some insights on the first as well. We could do worse than to turn to for an account of this.6 An Aristotelian view of character has the significant Conflicts between aspects of self-identity in different virtue of viewing identity as both real and consequential as contexts certainly do not arise for the first time with well as also being an object of work. We have on his view a social media, and are not limited to identities which are determinate character—e.g., we may, in fact, be a coward. discriminated against. One does not, for the most part, But in this view we still need not fall into Sartrean bad faith, discuss one’s sex life in church, even if that sex life takes for a coward need not be a coward in the sense that Sartre’s place within marriage—and within a straight marriage, and waiter is a waiter.7 A coward may be a coward, but may involves “vanilla sex” rather than BDSM, and so on. And yet nevertheless be brave in this or that particular situation— it is not without reason that recent years have seen renewed and through an accretion of such instances of bravery, may and intensified discussion of managing boundaries become brave rather than cowardly. Aristotle, along with between personal and professional life, and the tendency AA, tells us to “fake it ‘til you make it,” and both rightly of social media to either blur or overlap contexts of identity view this “faking it” as a creation of integrity, not a mere performance has created a new environment of identity demonstration of its absence. performance causing new requirements for thinking about and managing identities.10 On a correspondence theory of integrity, this self-conscious performance of a character which we do not possess In contemporary digital environments, we are frequently appears as false representation, but this makes sense interacting simultaneously with persons from different only when we assume a complete, settled, and coherent personal and social contexts. Our friends and followers in character. We say someone is “acting with integrity” when social networking sites (SNS) are promiscuously intermixed. she takes an action in accordance with her values and We have only a single profile in each, and we cannot principles, even or especially when it goes against her self- choose which profile items—gender identity, religious interest. Integrity, then, is not a degree of correspondence identity, former employers, name—are viewable to which between character and behavior, but between values connections or groups of connections in our network. and behavior. One can even act with integrity by going Nor can we choose to have different presentations for against one’s character, as in the case of the coward who different connections or groups: we may portray ourselves nonetheless stands up for what she believes in a dangerous differently in social or work contexts, but can choose only situation; the alcoholic entering recovery who affirms “Iam a single profile picture. There are work-arounds, of course, intemperate” and concludes “therefore I will not drink.”8 but they are onerous, difficult to maintain, and sometimes violate terms of service agreements requiring single The sort of identity relevant to integrity, then, is not personal accounts and real names. Even using built-in affordances identity in a philosophical sense (for the mere unity of intended to aid in maintaining contextual integrity,11 such as apperception is not a thing to which I can stay true), nor is private accounts (Twitter), friend lists (Facebook), or circles it one’s actual character or habits (for to reduce oneself to (Google+), is difficult and socially risky: difficult because one’s history and habits is bad faith, and acting according managing such affordances requires significant upkeep, to our habits could well lead us away from integrity if our curation, memory, and attention; risky because members habits are vicious). Instead, the relevant sort of identity must of groups of which we are members tend to have their be that with which we identify. Certainly we can recognize own separate interconnections, online or off, and effective that we have traits with which we do not identify, and the boundary enforcement must include knowledge of these process of personal growth is the process of changing interconnections and accurate prediction of information our character in order to bring it into accordance with the flows across them. If you wish to convince your parents that values we identify with. As Suler has argued, disinhibition you’ve quit Facebook, how far out in their social networks does not necessarily reveal some “truer self” that lies must you go in excluding friends from viewing your posts? “underneath” inhibitions; disinhibition may instead make Aunts and uncles? friends? Friends of friends of us unrecognizable to ourselves.9 Our inhibitions—at the family? Or, in maintaining separation of work and personal least the ones we value; which we identify with—are part of life, how are you to know whether a Facebook friend or

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 17 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

Twitter follower might know someone in your office well conversations take place within overlapping contexts. A enough to mention that “Oh, I know a co-worker of yours! public post absent identity markers, a popular video, Sounds like you have some serious HR issues . . .” Social for example, may receive a simple comment from an “in­ media is indeed connecting us more than ever before, but group” friend (e.g., “Too bad she’s straight!”), and through there are many significant silos the structural integrity of such interactions a potentially sensitive social context may which we wish to maintain. coalesce around all those participants and passive viewers present—and all this without the “in-group” friend having These social silos were previously maintained not only by any cues that she has broken down a silo. How are we to non-simultanous interactions with different groups and know which of a friend’s user-defined groups we are in, organizational contexts, but also by the mundane barriers and how they are organized? of time and space, missing in digital and especially in SNS environments. In our offline lives, when one is in church These effects are related to prior theorizations of Meyrowitz’s one is not also simultaneously in the office, in one’s tennis “middle region,” Papacharissi’s “publicly private and partner’s car on a family vacation, in one’s adult children’s privately public spaces,” and Marwick and boyd’s “context living rooms—and, similarly, when one is out on the town, it collapse.”13 What is perhaps most distinctive about this is not also simultaneously the morning after, next Monday particular case is the way these identity performances are at lunch break, and five years later while interviewing for a tied to unitary SNS profiles and take place within shifting new position. Digital media do not limit information flows and interlocking publicities rather than across a public/ through time and space the same ways as do physically private divide. We are not seeing the private leaking out into based interactions, and our ability to predict to where the public so much as we are seeing a variety of regional information may flow and how it may matter to others and publics overlaid upon one another. In this, we are called in other contexts—and to project that prediction indefinitely to account for our contextual identities in a new way: our into the future and in relation to concerns which our future selves are displayed, through both our actions as well as selves may have—is obviously insufficient to inoculate through others’ interactions with us, simultaneously before ourselves against the “privacy virus” that SNS presents.12 a multiplicity of audience with which we may identify in Worse still, in the absence of these mundane architectural different ways. barriers of time and space, and the social barriers to which they give rise, even our most thoughtful connections may This is the most peculiar challenge to integrity in an age not be able to accurately perceive and maintain the limits of social media: we can no longer work out our own idea on information flows which we seek to maintain. of how our values and commitments can harmonize into an integral self. Siloed identity performances allow us The co-worker who we run into at the gay bar, regardless of to perform those aspects of our identity, understood as his sexual orientation, must have overcome potential social that version of ourselves with which we identify, which barriers by being sufficiently comfortable with presence in a fit within one context and another context, variously and context and location where a sexualized same-gender gaze in sequence. We can be gay in one context, Muslim in is considered normal and proper rather than deviant. Given another, and a soldier in another still, and whether and to these mundane conditions, those who may bump into a what extent those identities can be integrated can largely co-worker at the gay bar—whether they be taking part in a be sequestered as an issue for our own moral introspection community of common self-identification, or whether they and self-labor. Once these identities must be performed be gay-friendly straights who are there to see a drag show, before a promiscuously intermixed set of audiences, or because it’s just the best place in town to go dancing— integrity in the sense of staying true to our values takes on can at least know that the other party has similarly passed a newfound publicity, for we can no longer gain acceptance through these social filters. Although it may not be known within groups merely by maintaining the local expectations by either party what has brought the other there, both are for values and behaviors within each group in turn, but “insiders” insofar as they have each met these conditions, instead must either (1) meet each and all local expectations and are thus aware that this knowledge of one another, globally, (2) argue before others for the coherence of these conditioned by this limited mode of access, ought to be identities when they vary from expectations particular treated as privileged information to be transmitted only to each group with which we identify, or (3) rebuild and selectively. maintain silos where time, space, and context no longer create them. By contrast, identification of sexual orientation through SNS profile data requires only a connection of any kind arising Indeed, so striking is this change that some have worried within any context in order to grant access to potentially whether we are losing our interiority altogether. sensitive information. But even without this self-disclosure, all contacts from all contexts are welcome in the virtual gay INTEGRITY AND THE “ORGANIZATION MAN” bar that may be overlaid on the SNS user’s page and feed. The worry that maintaining multiple profiles, and with them A vague work contact, made at a professional conference, multiple selves, reflects a lack of integrity is a Scylla in the is invited along to passively overhear conversations within anxieties of popular discourse about SNS to which there communities which he might never have been invited and is a corresponding Charybdis: the fear that an emerging might never have made himself a party to—even if a user, “let it all hang out” social will destroy the private self for example, posts news of gay marriage legal triumphs and altogether, and ring in a new age of conformity, where all vacation pictures with her partner only to a limited “close aspects of our lives become performances before (and by friends” list, her page nonetheless remains a venue in which implication for) others.

PAGE 18 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

There are, however, significant reasons to believe that, even that our judgment will become impaired by groupthink and if our lives become ubiquitously subject to surveillance diffusion of responsibility, and work out ways in which we and coveillance, this will not result in the exclusion of can avoid making poor judgments under that organizational expressions of marginalized identities or unpopular views.14 influence. Social networks may metaphorically provide that First, we see tendencies towards formation of social and more-sober friend who asks “are you sure you’re okay to informational echo chambers, resulting in increasingly drive?,” enabling our better judgment to gain a foothold. extreme views rather than an averaging-out to moderate and universally accepted views, as Sunstein has argued for and Organizations may then have a similar relation to our integrity documented at length.15 But, secondly, even insofar as we as does our character. Our character is formed by a history do not separate ourselves out into social and informational of actions and interactions, but we may not identify with “Daily Me”s becoming a virtual “city of ghettos,” the messy the actions that it brings us to habitually perform. When we and contentious digital spaces in which we are called to recognize our vices—e.g., intemperance—and seek to act account for the integration of our multiple selves may tend in accordance with our values and beliefs, we act against not only towards safe and “lowest-common denominator” our character and contribute thereby to reforming our habits versions of self-expression, but also towards greater and character to better align with the version of ourselves visibility and impact of divergent views, and even a new with which we identify. Organizations may similarly bring impetus away from conformity.16 us, through their own form of inertia and habituation, to act in ways contrary to our values and beliefs. A confrontation Thus far we have considered how limiting information with this contradiction through context collapse may help flows across social and organizational contexts can us to better recognize the organization’s vices and to act promote integrity, but it is certainly true as well that such according to the version of ourselves, in that organizational siloing of different self-performances can support a lack context, with which we identify—and contribute thereby to of integrity. Compartmentalization is a key tool in allowing reforming our organization to better align with our values, diffusion of responsibility. The employee who takes an and with its values as well. “I just work here” perspective in her professional life is more likely to encounter productive cognitive dissonance NOTES when participating in the mixed contexts of SNS in which 1. D. Kirkpatrick, The Facebook Effect, 199. discussions with co-workers about their employer’s actions 2. M. Zimmer, “Facebook’s Zuckerberg,” n.p. are subject to viewing and commentary by other friends who may view a corporate triumph as an environmental 3. K. Healy, “Actually, Having One Identity for Yourself Is a Breaching Experiment,” n.p. disaster. The churchgoer who has come to a private peace with her personal rejection of some sectarian dogmas may 4. B. Stone and S. Frier, “Facebook Turns 10,” n.p. be forced into a more vocal and public advocacy by having 5. D. Hume, A Treatise of Human Nature, I.4.6. to interact simultaneously with various and divergent 6. Aristotle, “Nicomachean Ethics,” 1729–1867. friends’ reactions to news of court rulings about 7. J.-P. Sartre, and Human Emotion; Sartre, Being and rights. Nothingness, 101–03. 8. To forestall a possible misunderstanding, I do not mean to claim In these sorts of cases, there is a clear threat to identity that alcoholism is a matter of character. As I understand it, the performances, placing users into precarious positions common view among those who identify as alcoholics is that it is wherein they must defend and attempt to reconcile a disease and a permanent condition—what is subject to change seemingly incompatible group identifications—but this loss is whether the alcoholic is keeping sober or has relapsed. This is where character comes into play—specifically, the hard work in the user’s tranquility, in some cases, may bring with it a of (re)gaining and maintaining the virtue of temperance through gain in personal integrity and possibilities for organizational abstemiousness. reform. While it is certainly a bad thing that intermixing 9. J. Suler, “The Online Disinhibition Effect.” of audiences may subject users to , and 10. Discussion in the first part of this section covers material separate performances of identities proper to different addressed more systematically in D. E. Wittkower, “Facebook groups and contexts need not be indicative of a lack of and Dramauthentic Identity.” integrity, compartmentalization can also enable people to 11. H. Nissenbaum, “Privacy as Contextual Integrity.” act against their own values and stifle productive criticism within organizations. 12. J. Grimmelmann, “The Privacy Virus.” 13. J. Meyrowitz, No Sense of Place; Z. Papacharissi, A Private Sphere; Luban et al. argue forcefully, with reference to the Milgram A. Marwick and D. Boyd, “I Tweet Honestly, I Tweet Passionately.” experiment, that bureaucracies create a loss of personal 14. S. Mann et al., “Sousveillance.” responsibility for collective outcomes, resulting in what 15. C. Sunstein, Republic.com 2.0; Sunstein, Going to Extremes. Arendt called “rule by nobody.”17 They suggest that 16. N. Negroponte, Being Digital; E. Pariser, The Filter Bubble; we should attempt to maintain adherence to our moral Wittkower, “Facebook and Dramauthentic Identity.” values—maintain our integrity in the sense of staying 17. D. Luban et al.; H. Arendt, On Violence, 38-39. true to the version of ourselves with which we identify— by analogy to how we think of our responsibility for our BIBLIOGRAPHY actions when under the influence of alcohol. Just as we Arendt, H. On Violence. New York: Harcourt, Brace & World, 1969. plan in advance for our impaired judgment later by taking a cab to the bar or designating a driver, so too, before we Aristotle. “Nicomachean Ethics.” In The Complete Works of Aristotle, edited by J. Barnes Princeton, NJ: Princeton University Press, 1984. enter into an organizational context we should be aware

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 19 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

Grimmelmann, J. “The Privacy Virus.” In Facebook and Philosophy, edited by D. E. Wittkower. Chicago: Open Court, 2010. The Moral Roots of Conceptual Confusion Goffman, E.The Presentation of Self in Everyday Life. New York: Doubleday, 1959. in Artificial Intelligence Research Healy, K. “Actually, Having One Identity for Yourself Is a Breaching Experiment.” Crooked Timber, May 14, 2010. Retrieved from http:// crookedtimber.org/2010/05/14/actually-having-one-identity-for­ Niklas Toivakainen yourself-is-a-breaching-experiment/. UNIVERSITY OF HELSINKI Hume, D. A Treatise of Human Nature. Project Gutenberg, 2012. Retrieved from http://www.gutenberg.org/files/4705/4705-h/4705-h. htm. INTRODUCTION Kant, I. Critique of Pure Reason. New York: Cambridge University Press, I gather that it would not be an overstatement to claim that 1998. the field of Artificial Intelligence (AI) research is perceived Kirkpatrick, D. The Facebook Effect. New York: Simon & Schuster, 2010. by many to be one of the most fascinating, inspiring, Luban, D., A. Strudler, and D. Wasserman. “ in the hopeful, but also one of the most worrisome and dangerous Age of Bureaucracy.” Michigan Law Review 90, no. 8 (1992): 2348–92. advancements of modern civilization. AI research and Mann, S., J. Nolan, and B. Wellman. “Sousveillance: Inventing and related fields such as neuroscience promise to replace Using Wearable Computing Devices for Data Collection in Surveillance human labor, to make it more efficient, to integrate robotics Environments.” Surveillance & Society 1, no. 3 (2003): 331–55. into social realities,1 and to enhance human capabilities. Marwick, A., and D. Boyd. “I Tweet Honestly, I Tweet Passionately: To many, AI represents or incarnates an important element Twitter Users, Context Collapse, and the Imagined Audience.” New Media & Society 13, no. 1 (2011): 114–33. of a new philosophy of mind, contributing to a revolution in our understanding of humans and life in general, which Meyrowitz, J. No Sense of Place: The Impact of Electronic Media on Social Behavior. New York: Oxford University Press, 1986. is usually integrated with a vision of a new era of human and super human intelligence. With such grandiose hopes Negroponte, N. Being Digital. New York: Vintage, 1996. invested in a project it is nut surprising that the same Nissenbaum, H. “Privacy As Contextual Integrity.” Washington Law Review 79, no. 1 (2004): 119–57. elements that invoke hope and enthusiasm in some, generate anxiety and disquietude in others.2 Papacharissi, Z. A Private Sphere: Democracy in a Digital Age. Malden, MA: Polity Press, 2010. While I will have things to say about features of these visions Pariser, E. The Filter Bubble: How the New Personalized Web Is Changing What We Read and How We Think. New York: Penguin, 2012. and already existing technologies and institutions, the main ambition of this paper is to discuss what I understand Sandel, M. “The Procedural Republic and the Unencumbered Self.” Political Theory 12, no. 1 (1984): 81–96. to be a pervasive moral dimension in AI research. To make my position clear from the start, I do not mean to Sartre, J.-P. Being and Nothingness. New York: Washington Square Press, 1993. say that I will discuss AI from a moral perspective, as if it could be discussed from other perspectives detached Sartre, J.-P. Existentialism and Human Emotion. New York: Citadel, 2000. from morals. I admit that thinking about morals in terms Stone, B., and S. Frier. “Facebook Turns 10: The Mark Zuckerberg Interview.” Business Week, January 30, 2014. Retrieved from http:// of a “perspective” is natural if one thinks of as www.businessweek.com/printer/articles/181135-facebook-turns-10­ corresponding to a theory about a separable and distinct the-mark-zuckerberg-interview. dimension or aspect of human life, and that there are Suler, J. “The Online Disinhibition Effect.”CyberPsychology & Behavior other dimensions or aspects, say, scientific reasoning for 7, no. 3 (2004): 321–26. instance, which are essentially amoral or “neutral” with Sunstein, C. Republic.com 2.0. Princeton, NJ: Princeton University respect to morality. Granting that it is a common trait of Press, 2009. modern analytical philosophy and scientific thinking to Sunstein, C. Going to Extremes: How Like Minds Unite and Divide. New precisely presuppose such a separation between fact and York: Oxford University Press, 2011. morality (or “value” as it is usually perceived), I am quite Wittkower, D. E. “Facebook and Dramauthentic Identity: A Post- aware that moral considerations enters into the discussion Goffmanian Model of Identity Performance on SNS.”First Monday, 19, of AI (as is the case for all modern techno-science) as a no. 4 (2014): n.p. Retrieved from http://firstmonday.org/ojs/index.php/ fm/article/view/4858/3875. distinct and separate consideration. Nevertheless, I will not be concerned here with a critique of moral evaluations Zimmer, M. “Facebook’s Zuckerberg: ‘Having Two Identities for Yourself Is an Example of a Lack of Integrity’,” May 5, 2014. Michaelzimmer.org. relevant for AI research—as, for instance, an ethics Retrieved from http://www.michaelzimmer.org/2010/05/14/facebooks­ committee would be—but rather with radicalizing the zuckerberg-having-two-identities-for-yourself-is-an-example-of-a-lack­ relationship between morality and techno-science.3 My of-integrity/. main claim in this paper will be that the project of AI—as the project of any human endeavor—is itself inextricably a moral matter. Much of what I will be doing here is to try and articulate how this claim makes itself seen on many different levels in AI research. This is what I mean by saying that I will discuss the moral dimensions of AI.

AI AND TECHNO-SCIENTIFIC UNDERSTANDING OF NATURE The term “Artificial Intelligence” invites three basic philosophical—i.e., conceptual—challenges: What is (the

PAGE 20 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

meaning of) “artificial” and “intelligence,” and what is philosophy, devising new methods and practices which the idea of these two coupled together? For instance, built on (very) different and aspirations. if one takes anything “artificial” to be categorically (conceptually, metaphysically?) distinct from anything It would take not one but many volumes to clarify all the “genuine,” “real,” or “natural”—which it conceptually different (trans)formative forces that led up to the birth seems to suggest—and if we think it sufficient (for a of the new methods and cosmology of modern techno­ given purpose) that “intelligence” be understood as a science, and many good books have been written on the computational/mechanical process of some sort, then subject.7 Nevertheless, I shall shortly try to summarize what any chess playing computer program, not to speak of seems to me—with regards to the topic of this paper— the new master in Jeopardy IBM’s “,”4 would be to be some of the decisive differences between modern perceived as a real and successful token of AI (with good science and its ancient and medieval predecessors. We future prospects for advancement) and would not invoke begin by noting that in the Aristotelian and scholastic any philosophical concerns in us. But as can be observed natural philosophy, knowing what a thing is, was (also and when looking at the diverse field of AI research, there essentially) to know its telos or purpose as it was revealed are many who do not think that chess playing computers through the Aristotelian four different causal forces, and or Jeopardy master Watson display “intelligence” in any especially the notion of “final cause.”8 Further, within this “real” sense; that “intelligence” is not simply a matter cosmological framework, “Nature” or “the natural” stood of computing power. Rather, they seem to think that for that which creates itself or that which is essential—and there is much more to the concept of “intelligence” and so that which is created by human hands is of a completely how it relates to the concept of (an actual human) life different order. Thirdly, both Plato and Aristotle had than machines like Watson encompass or display. In placed the purely theoretical or formal arts or knowledge other words, the dissatisfaction with what is perceived hierarchically above “practical” knowledge or know-how as a limited or narrow conception of intelligence (arguably reflecting the political and ideological power invites the need for philosophical reflection as to what structures of the ancient Greek society). On the other hand, “intelligence” really means. I will come back to the in the paradigm of modern science, knowing what a thing is concept of “intelligence,” but let us begin by considering is to know how that thing functions, how it is “constructed,” the role the term “artificial” plays in this debate and the how it can be controlled and manipulated, etc. Similarly, philosophical and ideological weight it carries with itself. in the modern era the concept of “Nature” or “the natural” loses its position as that which is essential and instead Suppose we were of the opinion that Watson’s alleged becomes more and more perceived as the raw material for “intelligence,” or any other so-called “Good Old Fashion man’s industriousness. So, in contrast to the Platonic and Artificial Intelligence,”5 does not satisfy essential features Aristotelian glorification of the purely theoretical or formal of intelligence of the “sort” human intelligence builds on arts/knowledge, the seventeenth-century philosophers and that “more” is needed, say, a body, , moral drew on a new vision “of the importance of uniting theoria agency, etc. We might think all of this and still think that AI with paraxis, a vision that grants new prominence to human systems can never become, out of conceptual necessity, agency and labor.”9 In other words, the modern natural anything more than technological devices or systems, philosophers and scientists sought a knowledge that would albeit very sophisticated and human or animal like ones: enable them to dominate natural phenomena. there will always, so to speak, be an essential difference between a simulation and a real or natural phenomena— This was the cornerstone of Francis Bacon’s scientific this is what the term “artificial” conceptually suggests. revolution. For Bacon, as for his followers—arguably the But as we are all aware, this standpoint is not shared by whole project of modern techno-science—the duty of all and especially not within the field of AI research and human power was to manipulate, change, and refine much of “naturalistic philosophy of mind,” as the advocates corporeal bodies, thus conceptualizing “knowledge” as the of what is usually termed “strong AI” hold that AI systems capacity to understand how this is done.10 Hence Bacon’s can indeed become “real” or “genuine” “autonomous,” famous term “ipsa scientia potestas est” or “knowledge “intelligent,” and even “conscious” beings.6 itself is power.” This same idea can also be found at the heart of the scientific self-understanding of the father That people can entertain visions and theories about of modern philosophy and modern dualism (which also AI systems one day becoming genuinely autonomous sets the basis for much of the philosophy and theory of conscious intelligent beings without feeling that they AI), namely, in Descartes’s articulations. In explaining the are committing elementary conceptual mistakes derives of the new era of natural philosophy and its methods from the somewhat dominant conception of the nature he proclaimed that they will “render ourselves the masters of concepts such as “artificiality,” “life,” and the “natural/ and possessors of nature.”11 genuine” deep at the heart of the modern techno­ scientifically informed self-understanding or worldview. Now the main point of this short and crude survey is to As most of us are aware, modern science developed into try and highlight that had the modern scientific paradigm its paradigmatic form during the seventeenth century, not been built on a unity between theoria and praxis and reflecting a sort of culmination point of huge social, the ideas of the duty of man to dominate over nature, we religious, and political changes. Seen from the perspective would not have read Bacon proclaiming that the artificial of scientific theory and method, the founders and does not differ from the natural either in form or in essence, visionaries of modern science turned against the ancient but only in the efficient.12 For, as in the new Baconian Greek and medieval scholastic “contemplative” natural model, when nature loses (ideologically) its position as

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 21 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

essential and becomes predominantly the raw material for theory is shown by the power of manipulation it produces— man’s industriousness, nature (and thus life) itself becomes it simultaneously developed, due to diverse reasons, a self- nothing apart from how man knows it or will someday be image of political and value (moral) neutrality: a science able to know it—and here “knowledge” is conceptualized for the sake of science itself.16 This meant that while the as that which gives power over phenomena. And even more measure of knowledge was directly related to utility, power to the point, had such decisive changes not happened, we of manipulation, and control17 it was thought that this would not be having a philosophical discussion about AI knowledge could be attained most efficiently and purely in its “strong” sense—i.e., in the sense that the “artificial” when potentially corrupt individual interests of utility or can gain the same ontological status as the “real” or other values were left outside the methods, theories, and “natural”: when such a conceptual change has been made, practices of science.18 This principle gives modern science when the universe is perceived as essentially in no way its specific, specialized, and differentiated function in different than an artifact or technological device, when the modern society: as the producer of “objective” techno­ cosmos is perceived to essentially be captured through scientific knowledge. techno-scientific knowledge, then the idea of an AI system as a genuinely autonomous conscious intelligent being One of the main reasons for calling scientific knowledge becomes a thought to entertain. “neutral” seems to be founded on an urge to detach it as much as possible from the “use” this knowledge is put As I have pointed out, this modern and Baconian idea is to: it can be “misused,” but this is not to be blamed on echoed in thinkers all the way from Descartes—whom the institution of science, for it (ideally) deals purely with perceived all bodily functions as essentially mechanical objective facts. The real problem, one often hears, is the and subject to technological manipulation/control13—to politico-economic power structures that pervert scientific modern “naturalist functionalists” (obviously denying knowledge in pursuit of corrupted ends. This is why we Descartes’s substance dualism) who advocate AI in its need political regulation, for we know that scientific “strong” sense and suggest that life and humans are knowledge has high potency for power and thus destruction “made of mindless robots [cells] and nothing else, no non­ or domination. This is why we need ethics committees and physical, nonrobotic ingredients at all.”14 Claiming such ethical regulations, because science itself is unable to an essential unity between nature and artifact obviously ethically determine its moral status and regulate its domain goes, so to speak, both ways: machines and artifacts are of action: it only deals itself with supposedly amoral essentially no different than nature or life, but the main objective facts. argument and emphasis is really that nature and life are essentially no different from artifacts. In other words, I am, of course, not indicating that scientists are morally I would claim, what is expressed here—in the modern indifferent to the work they do. I am simply pointing out techno-scientific understanding of phenomena—is the that as a scientist in the modern world, one’s personality idea that it is the artificial (i.e., human power) that is the as a scientist (dealing with scientific facts) is differentiated primary or the essential. I will characterize this ideologically from one’s moral self-understanding in any other sense based conception as a technological or techno-scientific than the alleged idea that science has an inherent value understanding of nature, life, and being. Now the claim in itself. Obviously, any scientist might bring her moral I will attempt to lay out is that such a technological self with them to work and into the laboratories, so the understanding is, in contrast to how it is usually perceived, split does not have to occur on this level. Instead, the not simply a question of neutral objective facts, but rather split finds itself at the core of the idea of the “neutral and an understanding or perspective that is highly morally objective” facts of science. So when a scientist discovers charged. In the last part of the paper I will try to articulate the mechanisms of, say, a hydrogen bomb, the mechanism in what sense (or perhaps a particular sense in which) this or the “fact of nature” is itself perceived as amoral—it is claim has a direct bearing on our conceptual understanding what it is neutrally and objectively: the objective fact is of AI. neither good nor evil, for such properties do not exist in a disenchanted, devalorized, and rationally understood IS TECHNO-SCIENTIFIC UNDERSTANDING nature; nature follows natural (amoral) that are subject 19 AMORAL? to contingent manipulation and utilization. The reason that I pose the question of techno-science’s One problem with such a stance relates to what I will call relation to morality is that there resides within the self- “the hypocrisy of modern science.” On a more fundamental understanding of modern techno-science an emphatic level I would challenge the very idea that scientific separation between fact and value (as it is usually termed). It knowledge of objective facts of nature/reality is itself may be added that modern science is by no means the only “neutral” with respect to morals. Now, to begin outlining institution in our modern culture that upholds such a belief what I mean by the “hypocrisy of modern science,” let me and practice. In addition to the institutional cornerstone start by noting that the dawn of modern science carries of modern secular societies—namely, the separation with itself a new, perhaps unprecedented, democratic between state and church—the society at large follows a principle of open accessibility.20 In addition to the Cartesian specialization and differentiation of tasks and authorities/ idea that “Good sense or Reason, is by nature equal in all hierarchies.15 Techno-science is one, albeit central, of men,”21 one might say that the democratic principle was these differentiated institutions. Now despite the fact engraved in the method itself, for it was the right methods that modern techno-science builds strongly on a kind of of modern science, not aristocratic or elite minds, that unity between theory and practice—the truth of a scientific were to produce true knowledge “as if by machinery.”22

PAGE 22 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

Hence the new and its methods—both Bacon’s committed to and engaged with weapons research.29 At and Descartes’s—were to put men on “an equal footing.”23 least since the Second World War up until, say, the late Although the democratization of knowledge was part of 1980s, military technology R&D relied mostly on direct the ideology of Bacon, Descartes, and the founders of The funding by the state, as state policy (at least in the United Royal Society, the concrete reality was and is a completely States) was dominated by what is usually called “spin-off” different story. As an example, the Royal Society, founded thinking. The term “spin-off” refers to the idea and belief in 1660, did not have a single female member before that through heavy funding of military R&D, the civilian and 1945. Nor has access to the scientific community ever commercial sectors will also benefit and develop. So, as been detached from individuals’ social backgrounds and it was perceived, as military R&D yielded new high-tech positions (class), economic possibilities, etc., not to speak devices and related knowledge; some of this knowledge of cultural and racial factors. There is also the issue of how and innovations would then “flow downstream” and find modern science is connected to forms of both economic its place in the civilian commercial markets (in appropriate and ecological exploitation: modern science with its form). This was, arguably, one of the main “legitimatizing” experimental basis is and has always been highly dependent reasons for the heavy numbers of scientists working on large investments and growing capital—capital which directly for military R&D. at least historically and in contemporary socio-economic realities builds on exploitation of both human as well as But this relationship has changed now (if it ever really was natural resources.24 Nevertheless, one might argue, such an accurate description). For instance, in 1960 the U.S. prejudices are more or less part of an unfortunate history, Department of Defense funded a third of all Scientific R&D in and today we are closer to the true democratic ideals of the Western world, whereas in 1992 it funded only a seventh science which have always been there: so we can still hold of it.30 Today this figure is even lower due to a change in on to a separation between fact and morals. the way military R&D relates to civil commercial markets. Whereas up until the 1980s military R&D was dominated All the same, there is another form of hypocrisy that finds by “spin-off” thinking, today it is possible to distinguish itself deep in the roots of modern science and alive and at least up to eight different ways in which military R&D well, if not even strengthened, even today. As both Bacon is connected to and interdependent with civil commercial and Descartes clearly noted, the new methods of modern markets, spanning from traditional “spin-off” to its opposite science were to make men “masters and possessors “spin-in.”31 The modern computer and supercomputer, for of nature.”25 But the new methods of science would not example, are tokens of traditional spin-off and “Defense come only to serve man in his domination over nature, procurement pull and commercial learning,” and the basic for the power that this new knowledge gave also served science that grew to become what we today know as the man in his domination over man.26 As one may quite Internet stems from “Shared infrastructure for defence easily observe when looking at the interconnectedness programs and emerging commercial industry.”32 The case of the foundations of modern science with political and of Deep Brain Stimulation (DBS), which is used to treat economic interests of the newly formed states of symptoms related to Parkinson’s disease and people Europe and the Americas, it becomes clear that the history fromessential tremor,33 and which falls under the of modern techno-science runs in line with modern military category of “Brain Machine Interfaces” and has its relevance industry and technologies of domination.27 For example, for AI research, will serve as another telling example of Galileo also used his own calculations of falling objects the complex and interconnected web of techno-science in order to calculate ammunition projectile trajectories, and the military industrial complex. Developed within the while Descartes’s analytical geometry very quickly became civilian sector, DBS and related knowledge and technology utilized for improvements of ballistics.28 And in contrast are perceived to be of high importance to military R&D. to the democratic spirit of modern science—which An official NATO report document from 2009 makes the perhaps can be said to have made some “progress”—the following observation: “From a military perspective, interconnectedness of techno-science and military and knowledge [neuroscientific knowledge]development weapons research and development (R&D) (and other should focus on three transitions: 1) from clinical and forms of exploitation/destruction) is still very tight. That patient applications to applications for healthy users, 2) is to say, while it is certainly true that modern techno­ from lab (or controlled) environments to the field, and 3) science is not in any sense original in its partnership from fundamental knowledge to operational applications.”34 and interdependence with military and weapons R&D, it I emphasized the third transitional phase suggested by nevertheless, in its conceptual and methodological strive to the document in order to highlight just how fundamental gain power over phenomena, has created unprecedented and to the point Bacon’s claim that “knowledge itself is means of destruction, domination, and oppression—and, power” is and what the unity between theory and practice we must not forget, means of construction and perhaps means in the modern scientific framework: techno­ even liberation. In other words, modern techno-science scientific knowledge of the kind derived, for example, has not exclusively built on or led to dreams of liberation from neuroscientific and cognitive science research not and diminishment of suffering (as it quite often rhetorically only lends itself but co-creates the interdependence promises) but, as one might put it, the complete opposite. between basic scientific research and the military industrial complex, and finds itself everywhere in between “spin-off” In 1975, the Stockholm International Peace Research and “spin-in” utilization. Institute’s annual books record that around 400,000 scientists, engineers, and technicians (roughly half of Until today, the majority of applied neuroscience the entire world’s scientific manpower at that time!) were research is aimed at assisting people who suffer

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 23 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

from a physical, perceptual or cognitive challenge may be completely “objective”—that is, valid—and and not at performance enhancement for healthy yet designed to serve a certain political interest . users. This situation opens up opportunities for . . the fact that their knowledge is goal-orientated spin-off and spin-in between advanced (military) does not mean it doesn’t work.39 Human System Interaction knowledge and the accomplishments in neurotechnology for Proctor’s point is to my mind quite correct and his patients.35 characterization of scientific “objectivity” as validity that “works”—something that enables one to manipulate and We should be reminded here that the military- control phenomena—is of course, in perfect agreement industrial complex is just one frontier that displays the with Bacon’s definition of scientific knowledge.40 The main interconnectedness of scientific “fundamental knowledge” lesson here, as far as I can see it, is that in an abstract and and end specific utilization (“the means constraint the detached sense it might seem as if scientific “objectivity” ends”36). Adding to this, we might just as well think of really could be politically and morally neutral (in its essence). the interconnectedness of basic scientific knowledge in Nevertheless, and this is my claim, the conceptual confusion agricultural research and the food markets,37 or scientific arises when we imagine that “objectivity” can in an absolute research of the human and other genomes and, for sense be “neutral” and amoral. Surely any given human example, the drug industry. But I take the case of military practice can be neutral and autonomous relative to specific R&D to suffice for the point I am making. issues/things, e.g., neutral to or autonomous with respect to prevailing political ideologies, by which we would mean * * * that one strives for a form of knowledge that does not fall victim to the prejudices of a specific ideology. This should Now despite the historical and ongoing (and increasing) nevertheless not lead us into thinking that we can detach connection between modern science and military R&D and “objectivity” from “knowledge” or “knowing”—as if we other exploitative forces, I am aware of the fact that this could understand what “objectivity” is independently of connection can be perceived to be contingent rather than what “knowing” something is. In this more pervasive sense essential—this is why I called the above a discussion of the objectivity is always dependent, as one might put it, on “hypocrisy” of modern science. In other words, one may knowing, while knowing itself is always a mode of life and claim that on an essential and conceptual level we might reflects what might be called a moral-existential stance or still hang on to the idea of science and its “fundamental attitude towards life. The mere fact that we choose to call knowledge” as “neutral”—although I find it somewhat something “knowledge” draws upon certain values, and worrisome that due to reasons described above, alarm bells more essentially, on a dynamics of aspirations that reflect aren’t going off more than they are. Part of the difficulty our stance towards our lives, towards other human beings, with coming to grips with the neutrality status of modern other forms of life, and “the world.” But the recognition that science is that the issue is connected on two different we have come to call some specific stance towards life and levels. On the one hand, the neutrality of science has been the world “knowledge” also includes the questions “Why integrated into its methods and to its whole ethos when do we know what we know and why don’t we know what we modern science struggled to gain freedom from church don’t know? What should we know and what shouldn’t we and state control since the seventeenth century.38 Related know? How might we know differently?”41 By this I mean to to this urge to form an institution free from the grips of say that such questions, moral by nature, are included in the religious and political power structures and domination, questions of “Why has this gained the status of knowledge?” neutrality with respect to value has become an important and “Why have we given this form of knowledge such a criterion of “objectivity”: only if the methods of science are position in our lives?” So the moral question we should free from the distorting, corrupting, and vulnerable values ask ourselves is what is the moral dynamics that has led of individual humans, can it be guided in a pure form by guiding concepts such as “domination,” “power,” “control,” the objective stance of rational reason. But, one might ask, “artificial,” “mechanization,” etc., to become constitutional is it really so that if science was not value free, and, more for (modern scientific) “knowledge”? importantly, if it was essentially morally charged by nature, it would be deprived of its “objectivity”? I am aware that many philosophers and theorists would object to the way I seem to be implying that To me it seems that “objectivity” is not at all dependent moral understanding is prior to scientific or theoretical on value neutrality in any absolute sense, or rather, not understanding, and not, as I gather many would claim, dependent on being amoral. Of course, this does not mean that all moral reasoning is itself a form of proto-theoretical that certain values, perceived by individuals owing up rationalization. My claim is in a sense the opposite, for I to, say, certain social norms and conventions, might not am suggesting that in order to understand what modern distort the scientific search for “objectivity,” not to speak science and its rationale is, we need to understand what of objectivity in other forms of knowing and understanding. lies, so to speak, behind the will to project a techno­ Obviously it might do so. The point is rather that “neutrality” scientific perspective on phenomena, on “intelligence,” and “objectivity” is not the same thing. “life,” the “universe,” and “being.” In other words, this is not a question that can be answered by means of modern Neutrality refers to whether a science takes a scientific inquiry for it is this very perspective or attitude stand; objectivity, to whether a science merits we are trying to clarify. So, despite the fact that theories certain claims to reliability. The two need not have of the hydrogen bomb led to successful applications, and anything to do with each other. Certain sciences can in this sense be said to be “objective,” I am claiming

PAGE 24 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

that this objectivity is not and cannot be detached from [L]arge parts of the cognitive science community the political and moral dimensions of a the will to build a realise that “true intelligence in natural and hydrogen bomb: from a will to power. Rather, it seems to (possibly) artificial systems presupposes three me that the “objectivity” of the facts of the hydrogen bomb crucial properties: are reflections or manifestations of will for such a bomb (power), for knowledge of the “facts” of, say, a hydrogen 1. The embodiment of the system; bomb shows itself as meaningful, as something worth our attention, only insofar as we are driven or aspire to search 2. Its situatedness in a physical and social for such a knowledge/power. In other words, my point is environment; that it is not a coincidence or a contingent fact that modern techno-science has devised means of, for instance, mass- 3. A prolonged epigenetic developmental process destruction. As has put it, through which increasingly more complex cognitive structures emerge in the system as Their [the institution of techno-science] a result of interactions with the physical and “application” is not the contingent and possible social environment”47 result of a prior theoretical content; it is already an “application,” an instrumental device, a technology. My understanding of the situation is that the new emerging Besides, no authority (instance) exists that would theories and practices are an outcome of a felt need to be different from this device and from the scientific conceptualize “intelligence” or in a manner that knowledge materializing in it that would decide more and more resembles how (true and paradigmatic) whether or not it should be “realized.”42 cognition and intelligence are intertwined with the life of an actual (human/living) being. That is to say, there seems ARTIFICIAL INTELLIGENCE OR ARTIFICIAL LIFE? to be a need to understand intelligence and cognition as My initial claim was that if there is to be any serious more and more integrated with both embodied and social discussion about AI in any other sense than what technical life itself—and not only understand cognition as an isolated improvements can be made in creating an “artificial” function of -manipulation alá GOFAI. To my mind “intelligence”—and thus holding a conceptual distinction this invites the question to what extent can “intelligence” between real/natural and artificial—then intelligence or be separated from the concept of “life”? Or, to put it “the mind” must be understood as technological. The another way, How “deep” into life must we go to find the discussion that followed was meant to suggest that (i) the foundations of intelligence? (modern) scientific worldviewis a technological (or techno­ scientific) understanding of the world, life, and of being and In order to try and clarify what I am aiming for with this (ii) that such an understanding is founded on an interest for question, let us connect the concept of “intelligence” utility, control, manipulation, and domination—for power— with that of “language.” Clearly, there might be a specific and finally that (iii) modern science is fundamentally and moment in a child’s life when a parent (or some other essentially morally charged, and strongly so with the moral person) distinctly hears the child utter its “first word”—a questions of power, control, and domination. sound that is recognizable as a specific word and used in a way that clearly indicates some degree of understanding Looking at the diversity of theories and of of how the word can be used in a certain context. But, of AI, one will quite quickly come to realize that AI research course, this “first word” is not a miracle in the sense that is always an interplay between, on the one hand, a before the utterance, the child was completely deprived of technological demand/challenge and aspiration, and, on language or that it now suddenly “has” language: it is rather the other hand, a conceptual challenge of clarifying the a kind of culmination point. Now the question we might meaning of “intelligence.” As the first wave of AI research, ask ourselves is whether there is any (developmental) part or “Good Old Fashion Artificial Intelligence” (GOFAI),43 of a child’s life—up until the point of the “first word” and built on the idea that high-level symbol manipulation beyond—that we could so to speak skip without the child alone could account for intelligence, and since the Turing losing its ability to utter its “first word” and to develop machine is a universal symbol manipulator, it was quite its ability to use language. I do not think that this is an “natural” to think that such a machine could one day empirical question. For what we would then have to assume become genuinely “intelligent.” Today the field of AI is in such a case is that the “first word” is not a culmination much more diverse in its thinking and theorizing about of all the interaction and learning that the child had gone “Intelligence,” and as far as I can see the reason for this is through prior to the utterance, and this would mean that we that people have felt dissatisfaction not only with the kind could, for instance, imagine a child that either came into of “intelligence” the “top-down” systems of GOFAI are able the world already equipped with a “developed” capacity to to simulate, but more so because people are suspicious use language or that we could imagine a child just skipping with how “intelligence” is conceptualized under the banner over a few months (I mean “metaphysically” skipping of GOFAI. Today there is talk about how cognition and “the over them: going straight from, say one month old to five mind” is essentially grounded in the body and in action,44 months old). But, we might note, in imagining this, we thus making “robotics” (the body of the AI system) an make use of the idea “already equipped with a developed essential part of AI systems. We also hear about “situated capacity to use language,” which all the same builds on cognition,” distributed or de-centralized cognition, and the idea that the development and training usually needed “the extended mind.”45 Instead of top-down GOFAI, many is somehow now miraculously endowed within this child. are advocating bottom-up “developmental” approaches.46 We may compare these thought-experiments with the

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 25 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

real case of a newborn child who immediately after birth really be detached from the whole process or development crawls to his/her mother’s breast, who stops screaming of life. What I have not said—and I have tried to make this when embraced, etc. Is this kind of, what one might call, clear—is that I think that modern science or a technological sympathetic responsiveness not constitutive of intelligence understanding of phenomena and life is invalid or “wrong,” and language: if this responsiveness was not there from if our criterion is, as it seems to be, utility or a form of the start—as constitutive of life itself—how could it ever be verification that is built on control over phenomena. We are established? And could we imagine such an event without all witnessing how well “it works” and left to its own logic, the prenatal life in the womb of the mother, all the internal so to speak, modern science will develop indefinitely—we and external stimuli, interaction, and communication that do not know the limits (if there is such) to human power. the fetus experiences during pregnancy? And what about the pre-fetal stages and conception itself—can these be left In this final part I want to try and illustrate how what I have out from the development of language and intelligence? been trying to say makes itself shown in the idea of strong AI. My main argument is that while I believe that the idea My point here is, of course, that from a certain perspective, of strong AI is more or less implicitly built into the modern we cannot separate intelligence (or language) from life techno-scientific paradigm (and is thus a logical unfolding itself. I say “a certain perspective” because everything of this paradigm), the rationale behind it is more ancient depends on what our question or interest is. But by the and in fact reflects a deep moral concern one might say looks of it, there seems to be a need within the field of AI belongs to a constitutive characteristic of the human being. research to get, so to speak, to the bottom of things, to Earlier I wrote that a strong strand within the modern a conception of intelligence that incorporates intelligence techno-scientific idea builds on a notion that machines as it is fleshed out in human and other forms of life in its and artifacts are no different than nature or life, but that totality—to make the artificial genuine. And if this is the the main argument and emphasis is really that nature and aim, then my claim would be that “intelligence” and “life” life are essentially no different from artifacts—that it is cannot be separated and that AI research must try to figure the artificial, human power which is taken as primary or out how to artificialize not only “intelligence” but also “life.” essential. Following this suggestion, my concern will now In other words, any idea of strong AI must understand life or be this: What is the dynamics behind the claim that human being, not only intelligence or “the mind,” technologically, beings or life itself is formal (since any given AI system for if it is not itself technological, then how could it be would be a formal system), and what kind of understanding made so? or conception of human beings does it build on as well as what it overlooks, denies, and even represses? In the beginning of this section, I said that AI research is always the interplay between technological aspirations and There are obviously logical and historical reasons why conceptual enquiry. Now I will add to this that AI is first drawing analogies between humans and machines is not and foremost driven by a technological aspiration and that only easy (in certain respects) but also tells us something the conceptual enquiry (clarification of what concepts like true. Namely, machines have more or less exclusively “life” and “intelligence” means or is) is only a means to been created to simulate human or animal “behavior” in fulfill this end. That is to say, the technological aspiration order to support, enhance, intensify, and replace human shapes the nature of the conceptual investigation: it has labor48 and capability49 and occasionally for the purpose of predefined the nature of the end result. What makes the entertainment. And since this is so, it is only logical that ultimate technological fulfillment of strong AI different from machines have had to build on some analogies to human its sibling genetic engineering is that whereas the latter physiology and cognitive capability. Nevertheless, there is must in its pursuit to control and dominate the genetic another part to the story—one might call it the other side foundations of life always take for granted life itself—it of the coin—of mechanization that I want to introduce with must rely on re-production of life: it can only dominate a the help of a quote from Lewis Mumford. given life—the former aspires in its domination to be an original creator or producer of “intelligence,” and, as I Descartes, in analyzing the physiology of the would claim, of “life.” human body, remarks that its functioning apart from the guidance of the will does not “appear at THE MORAL DYNAMICS OF THE CONCERN FOR all strange to those who are acquainted with the MECHANIZATION OF INTELLIGENCE AND LIFE variety of movements performed by the different automata, or moving machines fabricated by I have gone through some effort to make the claim that human industry. . . . Such persons will look upon AI—in its strong sense—presupposes a technological this body as a machine made by the hand of understanding of life and phenomena in general. Further, I God.” But the opposite process was also true: the have tried to make the case that modern science is strongly mechanization of human habits prepared the way driven by a technological perspective—a perspective of for mechanical imitations.50 knowledge to gain power over phenomena—and that it makes scant sense to detach morals (in an absolute sense) It is important to note that Mumford’s point is not to from such a perspective. Finally, I have suggested that the claim any logical priority to the mechanization of human pursuit of AI is determined to be a pursuit to construct an habits over theoretical mechanization of bodies and artificial model/simulation of intelligent life itself, since natural phenomena but, rather, to make a historical to the extent we hope to “construct” intelligence as it is observation as well as to highlight a conceptual point fleshed out in human and other forms of life, it cannot about “mechanization” and its relations to human social

PAGE 26 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

discipline, regimentation, and control.51 Building on what on; the appropriate and expected behavior follows formal I said earlier, I will take Mumford’s point to support my rules. But note, exactly because this is the case, so is its claim that to both theoretically and practically mechanize opposite. That is to say, because “appropriate” behavior is phenomena is always (also) to force or condition it into a grammatically tied to formal rules/expectations, so would specific form, to formalize phenomena in a specific way. As also “inappropriate” behavior be: to each appropriate Bacon explained the relation between natural phenomena response and act there are various ways of breaking and scientific inquiry, nature reveals her secrets “under them, ways which are derived from the “appropriate” constraint and vexed.” Although it is clear that Bacon ones and become “inappropriate” from the perspective of thought (as do his contemporary followers) that such a the “appropriate.” So, for instance, if I were to drink too method would reveal the “true” nature of phenomena, we many cocktails or suddenly start dancing passionately should note, or I would claim, that it was and still is the with someone’s wife or husband, these behaviors would method itself which was/is the primary or essential guiding be “inappropriate” exactly because there are “appropriate” force and thus nature or phenomena had/has to be forced ones that they go against. The same goes for anything we into a shape convenient to the demands and standards would call “informal” since the whole concept of “informal” of experiment52—this is why we speak of a “controlled grammatically presupposes its opposite, i.e., “formal,” research environment.” Similarly, my claim will be that meaning that we can be “informal” only in relation to what to theoretically as well as practically—in other words, is “formal,” or rather seen from the perspective of “formal.” ideologically—mechanize/formalize (human) life, (human) One could, for instance, say that at some time during behavior, (human) intelligence, (human) relationships, is the evening the atmosphere at the party became more itself to force or condition, so to speak, human nature into informal. One might say that both “formal” and “informal” a specific form: formalize in a specific way with specific are part of the same language game. In other words, underlying purposes. Now, as my claim has been, these one might think of a cocktail party as a social machine or underlying purposes are essentially something that must mechanism into which each participant enters and must be understood in moral-existential terms—they are the use his rational ability to “play” along with the determined “rationale” behind the scientific attitude to the world and or expected rules in relation to his own motivations, goals, not themselves “scientific objects.” To this I now add fears of social pressure, etc. that the underlying purposes cannot be detached from what (the meaning of) phenomena are transformed into We all know, of course, that the formal as well as any informal under the scientific and mechanizing methods—and this part of a cocktail party (or any other social institution) is a obviously invites the question whether any instance is a means to discipline, regulate, control, regiment, effectuate, development, a re-definition or a confusion, distortion, or make efficient, polite, tolerable, etc. the way in which perversion of our understanding. human relations are fleshed out: to have formal rules—and all the social conditioning that goes into making humans Obviously, this is a huge issue and one I cannot hope to “obey” these customs—is a way to moderate any political argue for to the extent that a good case could be made or ideological differences that people might have, to avoid for the understanding that I am advocating. Nevertheless, I or control embarrassing and painful encounters between shall attempt, by way of examples, to bring out a tentative people, and emotional, passionate, and spontaneous outlining of how this dynamics makes itself shown in reactions and communication, etc. In other words, a cocktail human relationships and interaction and how it relates to party is to force or condition human nature into a specific the idea of strong AI. formalized form: it is to mechanize human nature and her interpersonal relationships.53 The point to be made here is * * * that understanding the role that formalizing in this sense has, has to include a moral investigation into why human Some readers might at first be perplexed as to the character relations create difficulties that need to be managed at all, of the examples I intend to use and perhaps think them and what are the moral reactions that motivate to the kinds naïve and irrelevant. Nevertheless, I hope that by the end of formalizations that are exercised. of the paper, the choice of the examples will be more clear and seen to have substantial bearing on the issue at To make my point a bit more visible, think of a dinner hand. It might be added that the examples are designed to invitation. To begin with, we might imagine that the conceptually elaborate the issue brought up in Mumford’s invitation comes with the words “informal dress,” which quote above and to shed light on the dynamics of the idea indicates that the receiver might have had reason to expect that human intelligence and life are essentially mechanical that the dress code could have been formal, indicating that or formal. there is an underlying “formal” pressure in the relationship/ invitation. In fact, having “informal dress code” written on Think of a cocktail party at, say, the president’s residence. an invitation is already a formal feature of the apparently Such an event would be what we would call “formal,” and formal invitation. Just the same, the invitation might the reason for this is that the expectations on each person’s altogether lack any references to formalities and dress behavior are quite strict: well organized and controlled, codes, which might mean any of three things: (i) It might be highly determined (although obviously not in any “absolute that the receiver will automatically understand that this will sense), predictable, etc. One is, for instance, expected not be a formal dinner with some specific dress code (for the to drink too many cocktails, not to express one’s emotions invitation itself is formal). (ii) It might mean that they will or desires on the dance floor or otherwise too much, not understand—due to the context of the invitation—that it will to be impolite or too frank in one’s conversations, and so be an informal dinner, but that they might have had reason

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 27 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

still to expect that such invitations usually imply some form the methods of science. But this in itself does not say more of formality (a pressure to understand the relationship as than that by using scientific methods such and such can be formal). Needless to say though, both of these play on the attained: i.e., power over phenomena cannot be attained idea of a “code” that is either expected or not expected. (iii) through moral understanding or insight. The third possibility—which is in a sense radical, although a commonly known phenomenon—is simply that the whole I am by no means trying to undermine how much of idea/concept of formalities/informalities does not present our (social) lives follow formal codes and how much of itself. That is to say, the invitation itself is neither formal society and human behavior functions mechanically in one nor informal. If my friend, with whom I have an open and sense or another. It is certainly true that what holds for a loving relationship, invites me over for dinner, it would be cocktail party holds also for many other social phenomena very odd and indicative of a certain moral tension in our and institutions. And it is also true that any given social relationship or lack of understanding if I were to ask him if or interpersonal encounter carries with itself a load of I should dress formally or informally:54 our relationship is different formal aspects (e.g., what clothes one wears has in this sense and to this extent a-formal. And, one might always a social stamp on it). In fact, one might say that the say, it will stay a-formal to the extent no conflict or difficulty formal aspect of human life is deeply rooted in language arises between us, potentially leading us to adopt a code itself.56 Nevertheless, the crucial point is that any formal of formality in order to manage, avoid, control, etc. the features—which clothes one wears, what social situation or difficulty that has come between us. There is, so to speak, institution one finds oneself in—do not dominate or control nothing formal/mechanical as such about the relationship the human encounter as far as individuals are able to stay in or “behavior” and if an urge to formalize comes from either the openness that invites itself.57 Another way of putting it is inside or outside it transforms the relationship or way of that it is not the clothes one wears or the party one attends relating to it: it now becomes formalized/mechanized, i.e., that by itself is “formal.” Rather, the “formal” makes itself it has now been contextualized with a demand for control, known only as a response to the, quite often, unbearable regimentation, discipline, politeness, moderation, etc. What openness: driven by a desire to control, regiment, etc. the I take this to be pointing at is that (i) if a relationship does not moral, and I would add, constitutive bond that makes itself pose a relational and moral difficulty there will be no need, known in encounters between people and even between urge, or reason to formalize or mechanize the relationship. humans and other life-forms: the formal is a morally This means that the way we relate to each other in such dynamic response to the a-formal openness. cases is not determined by social collective identities or roles—at least not dominantly—but is rather characterized To summarize, my point is (i) that a technological perspective by an openness towards each other. (ii) This indicates that (i.e., strong AI58) is, so to speak, grammatically bound to mechanization or codification of human relationships and what I have now called formal or mechanical aspirations behavior is a reaction to certain phenomena over which towards life and interpersonal relationships; (ii) what I have one places a certain demand of regulation, control, etc. called the a-formal openness cannot, so to speak, itself be made formal/mechanical, but can obviously be mechanized So a cocktail party attendee does not obviously have to in the sense that the openness can be constrained and understand his or her relationship to other attendees in controlled; and (iii) an AI system can, within the bounds terms of formal/informal, although the social expectations of technological knowledge and resources, be created and pressures might do so. If an attendee meets a fellow and developed to function in any given social context in attendee openly, kindly, and lovingly, as opposed to ways that resemble (up to perfection) human behavior “politely” (“politely” being a formal way of relating to as it is fleshed out in formal terms. But perceiving such another, hence part of a “mechanism”), then there is no social behavior, i.e., formal relationships, as essential and mechanism or determined cause or course of action to sufficient for what it is to be a person who has a moral specify. Rather, such an encounter is characterized by an relation to other persons and life in general, is to overlook, openness (and to which extent it is open depends on the deny, suppress, or repress what bearing others have on us persons in the encounter), in which persons encounter each and we on them. other, at least relatively, independent of what their social collective identities prescribe to them, so to speak, as an A final example is probably in order, although I am quite I to a you. In such an openness, as far as it is understood aware that much of what I have been saying about the in this openness, there is no technological knowledge to a-formal openness of our relationships to others will remain be attained for, whereas technological understanding obscure and ambiguous—also, I must agree, partly because always includes a demand over (to control and dominate) articulating clearly the meaning of this is still outside the phenomena; in an (morally) open relationship or encounter reach of my (moral) capability. In her anthropological studies “we do not find the attitude to make something yield to of the effects of new technologies on our social realities our will.”55 This does not mean, of course, that we cannot and our self-conceptions, Sherry Turkle gives a striking impose a mechanical/technological perspective over story that illustrates something essential about what I have phenomena, and in this case on human relationships, and been trying to say. During a study-visit to Japan in the early that this wouldn’t give us scientifically useful information. 1990s, she came across a surprising phenomenon that she, The point is that if this is done, then it must exactly be rightly I would claim, connects directly with the growing understood as imposing a certain perspective, seeks positive attitude towards the introduction of sociable to determine means of domination, regulation, control: robots into our societies. Facing the disintegration of the power. So in this respect it is definitely correct to say that traditional lifestyles with large at the core, Japan’s scientifically valid knowledge reveals itself only through young generation had started facing questions as to what

PAGE 28 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

to do with their elderly parents and how to relate to them. “child,” etc., as they are understood in terms of collective This situation led to a perhaps surprising (and disturbing) normativity, and to what extent one is open to the other solution/innovation: instead of visiting their parents (as as an I to a you. To put it once more, the idea of strong AI they might have lived far away and time was scarce), some is, as one might put it, the flip side of the idea that one’s started sending actors to replace them. relationships to, for instance, one’s parents was and is only a matter of “a child” relating to “parents,” i.e., relating to The actors would visit and play their [the children’s] each other exclusively via collective social identities. parts. Some of the elderly parents had dementia and might not have known the difference. Most I am, of course, aware that anyone who will be advocating fascinating were reports about the parents who for strong AI will simply conclude that what I have called knew that they were being visited by actors. the a-formal openness of human relationship to others They took the actor’s visits as a of respect, and to life is something that must be “naturalized,” enjoyed the company, and played the game. When “disenchanted,” and shown to finally be formal/mechanical I expressed surprise at how satisfying this seemed in its essence. To this I cannot here say anything more. The for all concerned, I was told that in Japan being only thing that I can rely on is that the reader acknowledges elderly is a role, just as being a child is a role. the morally charged dimensions I have tried to articulate, Parental visits are, in large part, the acting out of which makes the simple point that understanding what it scripts. The Japanese valued the predictable visits means to place a technological and mechanical perspective and the well-trained courteous actors. But when I on phenomena always concerns a moral question as to heard of it, I thought, “If you are willing to send in what the demand for mechanization is a reaction to and an actor, why not send in a robot?”59 what it strives for. And obviously, my point has been that any AI system will be a formal system and is conceptually/ And, of course, a robot would, at least in a certain sense, grammatically bound to a technological perspective do just as well. In fact, we are not that far from this already and aspiration, which indicates, not that this sets some as the elderly-care institution is more and more starting to “metaphysical” obstacles for the creation of “strong AI,”60 replace humans with machines and elaborating visions of but rather that there is inherent confusion in such a fantasy future mechanization (and not only in Japan!)—as is, for in that it fails to acknowledge that it is a technological instance, also the parenting institution! It might be said that demand that is placed on phenomena or life.61 Turkle’s example, as it is in a sense driven to a quite explicit extreme, shows how interpersonal relationships, when CONCLUDING REMARKS dominated by formal codes and roles, hides or masks, I realize that it might not be fully clear to the reader how shuts out, suppresses, or even represses the a-formal open or in what sense this has bearing on the question of AI, encounter between individuals. As Turkle’s report illustrates, and especially on “strong AI.” To make it as straightforward what an actor, or robot for that matter, can do is to play the as possible, the central claim I am advocating for is that role of the child—and here “child” and “parent” are formal technological or mechanical artifacts, including AI systems, categories. What the actor (as an actor) cannot do is to be all stem from what I have called a “formal” (encompassing another person who responds to you and gives expression the “informal”) perspective on phenomena. And as this to, say, the fear of losing you. The actor (as an actor) might perspective is one that, as one might put it, contextualizes surely take on the role of someone responding/relating to phenomena with a demand for control, discipline someone, but that means that the actor would derive such regimentation, management, etc., and hence transforms feelings from, say, his/her own life and express them to it, it becomes an artifact of our demand. So my claim is you as another co-player/actor in the script that is being that the idea of strong AI is characterized by a conceptual played. In other words, the actor (as an actor) would not confusion. In a certain sense one might understand my claim relate to you as him/herself. If the actor, on the other hand, to be that strong AI is a logical/conceptual impossibility. would respond to you as him/herself, he or she would not And in a certain sense this would be a fair characterization, anymore be (in the role of) an actor, but would have to set for what I am claiming is that AI is conceptually bound to this aside. My claim is that a robot (AI system) could not what I called the “formal,” and thus always in interplay do this, that is, to set aside the part of acting upon formal with what I have called the a-formal aspect of life. So the scripts. What it can do is to be (play the role of) “a child” claim is not, for instance, that we lack a cognitive ability or or a “parent,” to the extent that these categories designate epistemic “perspective” on reality that makes the task of formal roles, but it could not be a being that is composed, strong AI impossible. The claim is that there is no thought so to speak, of the interplay or dynamics between the to be thought which would be such that it satisfied what we formal and the a-formal openness. And even though my want, urge for, or are tempted to fantasize about—or then or your culture might not understand parental relations as we are just thinking of AI systems as always technological formally as the Japanese in Turkle’s report, it is undeniable simulations of an non-technological nature. In this sense, that parent-child relationships (due to moral conflicts and the idea of strong AI is simply nonsense. But in contrast social pressure—just look at any psychoanalytical analysis!) to some philosophers coming from the Wittgenstein- take on a formal character—so there is no need to think influenced school of , I do not want that this is only a “Japanese phenomena.” One could, or to claim that the idea of “strong AI” is nonsense because rather should, say: it is a constant moral challenge and it is in conflict with some alleged “rules” of language or self-investigation to clarify how much of our relationship goes against the established conventions of meaningful to others (e.g., to one’s parents or children) is determined language use.62 Rather, the “nonsense” (which is to my or formed by the formal categories of, e.g., “parent,” mind also a potentially misleading way of phrasing it) is

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 29 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

a form of confusion arising out of a temptation or urge to 21. Descartes, The Philosophical Works of Descartes: Discourse on avoid acknowledging the moral dynamics of the “formal” the Method Part I, 81. and “a-formal,” of the openness inherent in our relationship 22. Bacon, Novum Organum, Preface, 7. to other and to life. It is a conceptual confusion but it is 23. Proctor, Value Free Science?, 26-27. moral by nature, which means that the confusion is not 24. Pereira, From Western Science to Liberation Technology; simply an intellectual mistake or shortcoming, but must be Mumford, Technics and Civilization. understood through a framework of moral dynamics. 25. Descartes, The Philosophical Works of Descartes: Discourse on the Method Part VI, 119. NOTES 26. C.f. Bacon, Novum Organum, 1.129, 62-63. Let me just note here 1. See Turkle, Alone Together. that I am certainly not implying that it is only modern science 2. See, for instance, Kurzweil, , and Malone, that serves and has served the cause of domination. This is “How Artificial Intelligence Will Shape Our Lives.” obviously not the case. My main claim is that, in contrast to at least ancient and medieval science, modern science builds 3. In this article I use the term ”techno-science” to characterize both conceptually as well as methodologically on a notion of the dominant self-understanding of modern science as such. In power. The consequence of this is and has been the creation of other words, I am claiming, for reasons which will become clear unprecedented means of domination (both in form of destruction —although not argued for sufficiently—that modern science is and opression as well as in construction and liberation!). predominantly a techno-science. I am quite sympathetic with

Michel Henry’s characterization, that when science isolates 27. Mumford, Technics and Civilization; von Wright, Vetenskapen itself from life as it is lived out in its sensible and interpersonal och Förnuftet; Taylor, A Secular Age; Mendelssohn, Science and nature—as modern science has done—it becomes a techno­ Western Domination. science. As Henry puts it, science alone is technology. See Henry, 28. Mendelssohn, Science and Western Domination, 77 & 207. Barbarism. For more on the issue, see, for instance, Ellul, The Technological Bluff; Mumford, Technics and Civilization; and von 29. Uberoi, The European Modernity, 90. Wright, Vetenskapen och Förnuftet. 30. Alic et al., Beyon Spinoff, 5. 4. See http://www-03.ibm.com/innovation/us/watson/. 31. Reverse spin-off or “spin-in.” Technology developed in the civil 5. See the short discussion of the term “Good Old Fashion Artificial and commercial sector flows upstream, so to speak, into military Intelligence” later in this article. uses. See ibid., 64–75. 6. Dennett, Consciousness Explained; Dennett, Sweet Dreams; 32. Ibid: 65-66, and 69. Haugeland, Artificial Intelligence: The Very Idea. 33. See http://www.parkinson.org/Parkinson-s-Disease/Treatment/ 7. See, for instance, Mumford, Technics and Civilization; Proctor, Surgical-Treatment-Options/Deep-Brain-Stimulation. Value Free Science?; Taylor, A Secular Age. 34. van Erp et. al., Brain Performance Enhancement for Military 8. In the Aristotelian system, natural phenomena had four “causal” Operations, 11-12. Emphasis added. forces: substance, formal, moving, and final cause. Proctor, Value Free Science?, 41. Of these causes, the moving or “efficient 35. Ibid., 11. cause” was the only one which remained as part of the modern 36. Proctor, Value Free Science?, 3. experimental scientific investigation of natural phenomena. Bacon, Novum Organum, II, 9, pp. 70. 37. For an interesting read on the effects of the inter-connectedness between scientific research and industrial agro-business in India, 9. Proctor, Value Free Science?, 6. see Kothari and Shrivastava, Churning the Earth. 10. Bacon, Novum Organum, 1. 124 pp. 60; Lång, Det Industrialiserade, 38. Taylor, A Secular Age; Proctor, Value Free Science? 96. 39. Proctor, Value Free Science?, 10. 11. Descartes, The Philosophical Works of Descartes, Discourse on Method part VI, 119. 40. Another example closer to the field of AI research would be ’s claim that the theoretical basis and methodological 12. Proctor, Value Free Science?, 22. tools used by him and his fellow champions of cognitive 13. See, for instance, Descartes’s Discourse on Method and Passions neuroscience and AI research are well justified because of of the Soul, in Descartes, The Philosophical Works of Descartes. the techno-scientific utility they produce. See Dennett, Sweet We might also note that , in addition to Dreams, 87. Descartes’s technological conception of the human body, gave a 41. Proctor, Value Free Science?, 13. technological account of the human soul, holding that cognition is essentially a computational process. Hobbes, Leviathan, 27­ 42. Henry, Barbarism, 54. Emphasis added. 28. See also Haugeland, Artificial Intelligence, 22. 43. Or top-down AI, which is usually referred to as “Good Old 14. Dennett, Sweet Dreams, 3. See also Dennett, Consciousness Fashion Artificial Intelligence” (GOFAI). See Haugeland, Artificial Explained; and Haugeland, Artificial Intelligence. Intelligence. 15. Habermas, The Theory of Communicative Action, Vol. 1 and Vol. 2; 44. Barsalou, Grounded Cognition. Taylor, A Secular Age. 45. Clark, “Reasons, Robots, and the Extended Mind”; Clark, 16. C.f. Henry, Barbarism, chapter 3 “Science Alone: Technology.” Supersizing the Mind; Wilson, “Six Views of .” 17. As Bacon put it: truth and utility are the same thing. Bacon, 46. Oudeyer et. al., “Intrinsic Motivation Systems for Autonomous Novum Organum, I,124, 60. Mental Development.” 18. Proctor, Value Free Science?, 31-32. 47. Guerin, 2008, 3. 19. One of the main ideological components of modern secularized 48. A telling example is, of course, the word “robot,” which comes techno-science has been to devise theories and models of from the Check “robota,” meaning “forced labor.” explanation that devalorized the world or nature itself. Morals are a human and social “construct.” See Proctor, Value Free Science?, 49. AI seen purely as a form of technology, without any philosophical and Taylor, A Secular Age. or metaphysical aspirations, falls under at least three different categories: (i) compensatory, (ii) enhancing, and (iii) therapeutic. 20. von Wright, Vetenskapen och Förnuftet, 53; Robinson, Philosophy For more on the issue, see Toivakainen, “Man and His Invention”; and Mystification. and Lin et. al., .

PAGE 30 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

50. Mumford, Technics and Civilization, 41. Emphasis added. Descartes, Rene. The Philosophical Works of Descartes, 4th ed., translated and edited by Elizabeth S. Haldane and G. R. T. Ross. New 51. Sherry Turkle gives contemporary examples of this logic York: Cambridge University Press, 1967. that Mumford is highlighting. Based on her fieldwork as an anthropologist, she has noted that sociable robots become Ellul, Jacques. The Technological Bluff, trans. W. Geoffery Bromiley. either possible or even welcomed replacements for humans Grand Rapids, Michigan: W. B. Eerdmans Publishing Company, 1990. when the context of human relationships into which the robots Habermas, Jürgen. The Theory of Communicative Action, Vol. 1, Reason are designed enter is mechanized and regimented sufficiently. and the Rationalization of Society. London: Heineman, 1984. For example, when a nurse’s job has become sufficiently mechanized/formal (due to resource constraints), the idea of a ———. The Theory of Communicative Action, Vol. 2, Lifeworld and robot replacing the nurse enters the picture. See Turkle, Alone System: A Critique of Functionalist Reason. Boston: Beacon Press, 1987. Together, 107. Hacker, P. M. S. Wittgenstein: Meaning and Mind, Volume 3 of an 52. In the same spirit, the Royal Society also claimed that the scientist Analytical Commentary on the Philosophical Investigations. Oxford: must subdue nature and bring her under full submission and Blackwell, 1990. control. von Wright, Vetenskapen och Förnuftet, 65. Haugeland, John. Artificial Intelligence: The Very Idea.Cambridge, MA: 53. For an interesting discussion of the conceptual and historical The MIT Press, 1986. relationship between mechanization and regimentation, Henry, Michel. Barbarism, translated by Scott Davidson. Chennai, India: discipline, and control of human habits, see Mumford, Technics Continuum, 2012. and Civilization. Hobbes, Thomas. Leviathan, edited by Ian Shapiro. New Haven, CT: Yale 54. Obviously I am thinking here of a situation in which my friend has University Press, 2010. not let me know that the dinner will somehow be exceptional, with perhaps an “important” guest joining us. Kenny, Anthony. Wittgenstein (revised edition) Oxford: Blackwell Publishing, 2006. 55. Nykänen, “Music and the Frailness of Wonder,” 130. Kothari, Ashish, and Aseem Shrivastava. Churning the Earth. New Delhi, 56. C.f. Wittgenstein, Philosophical Investigations, § 111. India: Viking, 2012. 57. For more on this issue see Backström, The Fear of Openness. Kurzweil, Ray. The Singularity Is Near: When humans Transcend Biology. New York: Viking, 2005. 58. Let me note here that the so called “weak AI” is not free from conceptual confusion either. Essentially a product of modern Lin, Patrick et. al. Robot Ethics. Cambridge, MA: The MIT Press, 2012. techno-science, it must also deal with the conceptual issue of Lång, Fredrik. Det Industrialiserade, Helsinki: Helsingin Yliopistopaino, how to relate questions of moral self-understanding with the 1986. idea of “knowledge as power” and “neutral objectivity.” Malone, Matthew. “How Artificial Intelligence Will Shape Our Lives.” 59. Turkle, Alone Together, 74. Emphasis added. ZDNet, July 19, 2012. http://www.smartplanet.com/blog/pure-genius/ 60. My point is, for instance, not to make any claims about the how-artificial-intelligence-will-shape-our-lives/8376, accessed October existence or non-existence of “qualia” in humans or AI systems 2013. for that matter. As far as I can see, the whole discussion about Mendelssohn, Kurt. Science and Western Domination. London: Thames qualia is founded on confusion about the relationship between & Hudson, 1976. the so-called “inner” and “outer.” Obviously, I will not be able to give my claim any bearing, but the point is just to encourage Mumford, Lewis. Technics and Civilization, 4th ed. with a new foreword the reader to try and see how the question of strong AI does not by Langdon Winner. Chicago: University of Chicago Press, 2010. need any discussion about qualia. Nykänen, Hannes. “Music and the Frailness of Wonder.” In Economic 61. I just want to make a quick note here as to the development within Value and Ways of Life, edited by Ralf Ericksson and Markus Jäntti. UK: AI research that envisions a merging of humans and technology. Avebury, 1995. In other words, . See Kurzweil, The Singularity Is Near; Oudeyer, Pierre-Yves et. al. “Intrinsic Motivation Sytems for Autonomous and www.kevinwarrick.com. If strong AI is to make any sense, Mental Development.” IEEE Transactions on Evolutionary Computation then this is what it might mean, namely, that humans transform 11, no. 2 (2007): 265–86. themselves to become “artificial” as far as possible (and we do not know the limits here!). Two central points to this: (i) A Pereira, Winin. From Western Science to Liberation Technology, 4th ed. will, just as genetic manipulation, always have to presuppose the Kolkata, India: Earth Books, 2006. givenness of life; (ii) cyborgs are an excellent example of human Proctor, Robert. Value Free Science? Cambridge, MA: Harvard University social and bodily life becoming (ideally fully?) technological. Press, 1991. The reason why the case of cyborgs is so interesting is that, as far as I can see, it really captures what strong AI is all about: to Robinson, Guy. Philosophy and Mystification. London: Routledge, 1997. not only imagine ourselves, but also to transform ourselves into Taylor, Charles. A Secular Age. Cambridge: The Belknap Press of Harvard technological beings. University Press, 2007. 62. C.f. Hacker, Wittgenstein: Meaning and Mind; Kenny, Wittgenstein. Toivakainen, Niklas. “Man and His Invention.” Njohja 3 (2014): 25–40. Turkle, Sherry. Alone Together. New York: Basic Books, 2011. BIBLIOGRAPHY Wilson, Margaret. “Six Views of Embodied Cognition.” Psychonomic Alic, John A. et al. Beyon Spinoff. Harvard Business School Press, 1992. Bulletin and Review 9, no. 4 (2002): 625–36. Backström, Joel. The Fear of Openness. Åbo University Press, Åbo, 2007. Wittgenstein, Ludwig. Philosophical Investigations, 3rd ed. Translated Bacon, Francis. Novum Organum. Memphis: Bottom of the Hill by G.E. M. Anscombe. New Jersey: Prentice Hall, 1953. Publishing, 2012. von Wright, G. H. Vetenskapen och Förnuftet, Vetenskapen och Barsalou, Lawrence. L. Grounded Cognition. In Annu. Rev. Psychol. 59 Förnuftet. Stockholm: Månpocket, 1986. (2008): 617–45. Uberoi, J. P. S. The European Modernity. New Delhi: Oxford University Clark, Andy. “Reasons, Robots, and the Extended Mind (Rationality for Press, 2002. the New Millenium).” Mind and Language 16, no. 2 (2001): 121–45. van der Zant, Tijn. et. al. (2013) “Generative Artificial Intelligence.” ———. Supersizing the Mind. New York: Oxford University Press, 2008. In Philosophy and Theory of Artificial Intelligence, edited by Vincent Müller. Berlin: Springer-Verlag, 2013. Dennett, Daniel. Consciousness Explained. Boston: Little, Brown and Company, 1991. van Erp, Jan B. F. et al. “Brain Performance Enhancement for Military Operations.” TNO: Netherlands Organisation for Applied Scientific ———. Sweet Dreams. Cambridge, MA: The MIT Press, 2006. Research, 2009. http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA567925, accessed September 10, 2013.

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 31 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

Hamlets in a thousand people’s eyes,” Luciano Floridi Increase or Decrease of Entropy: To mentioned that has different meanings in the beholders of different disciplines.6 His fundamental Construct a More Universal Macroethics principles of information ethics are committed to constructing an extremely metaphysical theory upon which (A Discussion of Luciano Floridi’s The could be grounded from a philosophical point of view. In a macroethical dimension, Floridi drew on Ethics of Information) his theories of philosophy of information, the “philosophia prima,” and constructed a non-standard ethics aliened from Xiaohong Wang any excessive emphasis on specific technologies, without DEPARTMENT OF PHILOSOPHY/INTERNATIONAL CENTER FOR looking into the specific behavior norms. PHILOSOPHY OF INFORMATION, XI’AN JIAOTONG UNIVERSITY, P.R. CHINA The four ethical principles of IE are quoted from this book as follows: Jian Wang DEPARTMENT OF PHILOSOPHY/INTERNATIONAL CENTER FOR 0 entropy ought not to be caused in the infosphere (null PHILOSOPHY OF INFORMATION, XI’AN JIAOTONG UNIVERSITY, law) P.R. CHINA 1 entropy ought to be prevented in the infosphere Kun Zhao SCHOOL OF ELECTRONIC AND INFORMATION ENGINEERING, 2 entropy ought to be removed from the infosphere XI’AN JIAOTONG UNIVERSITY, P.R. CHINA 3 the flourishing of informational entities as well as of the Chaolin Wang whole infosphere ought to be promoted by preserving, SCHOOL OF FOREIGN STUDIES, XI’AN JIAOTONG UNIVERSITY, cultivating, and enriching their well-being P.R. CHINA Entropy plays a central role in the fundamental IE principles ICTs are radically transforming our understanding of “self­ laid out by Floridi above, and through finding a more conception,” “mutual interactions,” “conception of reality,” fundamental and universal platform of evaluation, that is, and “interaction with reality,”1 which are concentrations through evaluating decrease or increase of entropy, he of ethics researchers. The timing is never more perfect commits to promote IE to be a more universal macroethics. to thoroughly rethink the philosophical foundations of However, as Floridi admitted, the concept of “entropy” information ethics. This paper will discuss Luciano Floridi’s that he has been using for more than a decade has indeed The Ethics of Information,2 particularly on the fundamental led to endless misconceptions and misunderstandings concepts of his information ethics (IE), the framework of the concept of “entropy” in thermodynamics and of this book and its implications on the Chinese IE, and Shannon’s information theory. Then, how can we solve the Floridi’s IE in relation to Chinese philosophical . alleged contradiction or divergence of Floridi’s concept of “entropy” (or metaphysical entropy) from the informational 1. THE BOOK FULFILLS THE HOPE IN and the thermodynamic concept of entropy? We think, as a “INFORMATION ETHICS, THE SECOND matter of fact, that the concept of entropy used by Floridi is equal to the latter two concepts, rather than not equal GENERATION” BY ROGERSON AND BYNUM. to them, though strictly relating to, as claimed by Floridi.7 In 1996, Simon Rogerson and Terrell Ward Bynum The key is to differentiate the informational potentiality coauthored an article “Information Ethics, the Second (informational entropy) from the informational semantic Generation.”3 They suggested that computer ethics, as the meaning (informational content). first-generation information ethics, was quite limited in research breadth and profundity, for it merely accounted for As Floridi explicitly interpreted, entropy in Shannon’s sense certain computer phenomena without a strong foundation can be a measure of the informational potentiality of an of ethical theories. As a result, it failed to provide a information source, “that is, its informational entropy.”8 comprehensive approach and solution to ethical problems According to this interpretation, in a system bearing energy regarding information and communication technologies, or information, the higher the entropy is, the greater the information systems, etc. For this, Luciano Floridi claims disorder and randomness are; and consequently, the more that far from being as it may deceptively seem at first sight, possibilities for messages being potentially organized in CE is primarily an ethics of being rather than of becoming, the system you have. Suppose in a situation of maximized and by adopting a level of abstraction the ontology of disorder (highest entropy), a receiver will not be able to CE becomes informational.4 Here we also refer to a vivid recognize any definite informational contents but nothing; analogy: a computer is a machine just as a washing however, nothing can mean everything, when people say machine is a machine, yet no one would ever conceive “nothing is impossible” or “everything is possible,” that is, the study of washing machine ethics.5 From this point of nothing contains every possibilities. In short, high entropy view, the prevalence of computer ethics resulted from means high possibilities of information-producing but some possible abuse or misuse. It’s therefore necessary to low explicitness of informational semantic meaning of an develop a paradigm for a second-generation information information source (the object being investigated). ethics. However, as the saying goes, “there are a thousand

PAGE 32 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

Though higher degree of entropy in a system means For example, Einstein’s theory of general relativity implied more informational potentiality (higher informational that rays of light should bend as they pass close to massive entropy ), a receiver could recognize less informational objects such as the sun. This prediction was a small content (the semantic meaning of information) at the time, probability event for those physicists living in the Newtonian making it difficult to decide what exactly the information paradigm, so are for common people living on the earth. So is. Inversely, the lower degree of entropy in a system “dark clouds” had been haunting in the sky of the classic means less informational potentiality (lower informational physics up until Einstein’s prediction was borne out by entropy) and less degree of randomness, yet a receiver Edington’s observation in 1919. Another classical case is could retrieve more informational content (the semantic in the history of chemistry, when Avogadro’s hypothesis meaning of information) at the time, making it less difficult was originally proposed in 1910. This argument was a small to decide what the exact information is. Given the above, probability event in the background of chemical knowledge Floridi set the starting point of four IE ethical principles: to at that time, and as a result, few chemists paid attention prevent from or remove increase of entropy. Or we revise to his distinction between atom and molecule so that the it a little and remain “to remove increase of entropy.” confronting situation among chemists had lasted almost From this point of view, we can say that Floridi’s concept for fifty years. As an example of that disorder situation, of entropy has entirely the same meaning as the concept Kekule gave as many as nineteen different formulas used of entropy in thermodynamics and Shannon’s information by chemists for acetic acid. This disorder finally ended theory. Entropy is a loss of certainty, comparatively; evil is after Cannizarro successful revived this hypothesis based a “privation of good.”9 on accumulated powerful experimental facts in the 1960s.

From Shannon’s information theory, “the entropy, H, of A period with high informational entropy is necessary for the a discrete random variable X is a measure of the amount development of science, in which scientific advancement is of uncertainty associated with the value of X,”10 and he incubated. Only after statements of such small probability explicitly explained an inverse relation between value of events are confirmed, however—and small probability entropy and our uncertainty of outcome output from an events change to be high probability events—can science information source: enter a stable and mature period. Only during this time can scientists solve problems effectively. As a result, each

H = 0 if and only if all the Pi but one are zero, this progressive step in science must be accompanied by a one having the value unity. Thus only when we are decrease of informational entropy of the objects being certain of the outcome does H vanish. Otherwise H investigated. Comparatively, information receivers need is positive.11 And, with equally likely events there is to remove increase of entropy in an information source in more choice, or uncertainty, when there are more order to have definite knowledge of the source. possible events.12 Floridi agrees with Weiner’s view, the latter thought that A philosophical sense of interpretation of Shannon’s entropy is “the greatest natural evil,”14 for it poses a threat mathematical formula runs as follows: to any object of possible values. Thus, the unnecessary increase of entropy is an irrational action creating evil. The amount of information I in an individual Inversely, any action maintaining or increasing information

message x is given by: I(x) = −log px is good. Floridi therefore believes any object or structure either maintaining or increasing information has at least This formula can be interpreted as the inverse of a minimum worth. In other words, the minimal degree of the Boltzmann entropy, and by which one of our moral value of inforgs could be measured by the fact that basic intuitions about information covered is: “any change may be morally good or bad not because of its consequences, motives, universality, or virtuous nature,

If px = 1 then I(x) = 0. If we are certain to get a but because the infosphere and the informational entities message it literally contains no ‘news’ at all. The inhabiting it are affected by it positively or negatively.”15 In lower the probability of the message is, the more this sense, information ethics specifies values associated information it contains.13 with , deontologism, , and . Speaking of his researches in IE, Floridi Let’s further the discussion by combing the explanation explained the IE “looks at ethical problems from the above with the informational entropy. When the potentiality perspective of the receiver of the action, not from the for information-producing is high (high informational source of the action, where the receiver of the action could entropy) in an information source, the occurrence of each be a biological or a non-biological entity. It is an attempt event is a small probability event on average, and a to develop environmental and ecological thinking one of the small probability event is informative (Popper’s high step further, beyond the biocentric concern, to develop an degree of falsification with ruling out many other logical ontocentric ethics based on the concept of what I call the possibilities). More careful thinking reveals, however, that infosphere. A more minimalist ethics based on existence, before the statement of such a small probability event can rather than on life.”16 Such a sphere combines the biosphere be confirmed, information receivers will be in a disordering and the digital infosphere. It could also be defined as an and confusing period of understanding the information ecosphere, a core ecological concept envisioned by Floridi. source, similar to the period of anomalies and crisis in the Within the sphere, the life of a human as an advanced history of science argued by Kuhn. Scientists under this intelligent animal is an onlife, a “Faktizitaet des Lebens” by disorder and confusion cannot solve problems effectively. Heidegger, rather than a concept associated with senses

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 33 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

and supersenses or transcendental dialectics. From this certain place in ethics, which could be interpreted from the perspective, Floridi’s information ethics actually lay a axiological analysis of information and the discussions on theoretical foundation for the first-generation computer five topics. ethics in a metaphysical dimension, fulfilling what Simon Rogerson and Terrell Ward Bynum hope for. The seventh chapter dwells upon the ethical problems of artificial intelligence, a focal point in current information 2. THE BOOK DEMONSTRATES ACADEMIC ethics studies. The eighth chapter elaborates upon the IMPORTANCE AND MAIN FEATURES AS FOLLOWS: constructionist values of Homo Poieticus. The ninth and tenth chapters explore the permanent topics of evil and IE is an original concentrate of Floridi’s past studies, a good. sequel to his three serial publications on philosophy of information, and an even bigger contribution to The eleventh chapter puts the perspective back on the philosophical foundation of information theories. In human beings in reality. Through Plato’s famous analogy of the book, he systematically constructed IE theories and the chariot, a question is introduced: What is it that keeps a elaborated on numerous information ethical problems from self a whole and consistent entity? Regarding egology and philosophical perspectives. Those fundamental problems its two branches and the reconciling hypothesis: the three are far-reaching, covering nearly all issues key to ethical life membranes model, the author provided an informational in an information society from an interdisciplinary approach. individualization theory of selves and supported a very The author cited rich references and employed detailed Spinozian viewpoint: a self is taken as a terminus of materials and meticulous analysis to demonstrate a new information structures growth from the perspective of field, which is created by information and ethics across their informational structural realism. related disciplines. They include ethical problems meriting immediate attention or long-term commitment based on The twelfth and thirteenth chapters seriously look into the author’s illustration of IE era and evolution, IE methods, the individual’s ethical issues that demand immediate and its nature and disciplinary foundations. In particular, the solutions in an information era, on the basis of preceding book constructs a unique framework with clear logic, well- self-theories. structured contents, and interconnected flow of thoughts from the beginning to the end, demonstrating the author’s In the fourteenth and fifteenth chapters, the IE problems strong scholarly commitment. in the economic globalization context are analyzed philosophically from an expanded point of view. General as The first chapter studies the ethics construction drawing it appears, it is thought-provoking. on the previously described information turn, i.e., the fourth turn. The pre-information turn era and the text code In the last chapter, Floridi neutrally discussed twenty era are re-localized with the assaults of information and critical views with humility, tolerance, and meticulousness, communication technologies. The global infosphere is and demonstrated his academic prudence and dedicated created, i.e., the informational generation of an ecological thinking. The exceptionally productive contention of system. It’s in fact a philosophical study of infosphere and different ideas will undoubtedly be even more distinct in inforgs transformation. his following works.

The second chapter gives a step-by-step examination 3. THE BOOK COMPRISES THREE and definition of the unified model of information ethics, INTERCONNECTED PARTS AS FOLLOWS: including informational resources, products, environment, and macroethics. It’s not difficult to see from the flow of thoughts in the book that IE, as the sequel to The Philosophy of Information,17 The third chapter illustrates the level of abstract (LoA) in is impressively abstract and universal on one hand and epistemology to clarify the interconnection of abstractness metaphysically constructed on information by Floridi with ontological commitments by taking telepresence as on another hand. In The Philosophy of Information, an example. he argued the philosophy of information covered a) the critical investigation of the conceptual nature and The following chapter presents a non-standard ethical basic principles of information, including its dynamics, approach in which the macroethics fosters a being-centered utilization, and sciences; b) the elaboration and application and patient-oriented information ethics, impacted by of information-theoretic and computational methodologies information and communication technologies and ethical to philosophical problems.18 The “theory plus application” issues. approach is extended in the book and constructed in an even succinct and clarified fashion. All in all, the first five The fifth chapter demonstrates that computer ethics is not chapters of the book define information ethics from a macro a discipline in a true sense. Instead, it’s a methodology and disciplinary perspective; the sixth to eleventh chapters and an . CE could be grounded upon IE examine the fundamental and everlasting questions on perspectives. information ethics. From the twelfth chapter onward, problems on information ethics are studied on individual, The sixth chapter illustrates the basic stance of information social, and global levels, which inarguably builds tiers and ethics, that is, the of the infosphere. In an strong logic flow throughout the book. object-oriented ethical model, information occupies a

PAGE 34 SPRING 2015 | VOLUME 14 | NUMBER 2 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

As a matter of fact, Floridi presents an even more profound the situation of IE studies in contemporary China is very approach in the design of research frameworks in the similar to that of the western IE studies before the mid­ book. The first five chapters draw on his past studies 1990s: There had been little multi-disciplinary work and on information phenomena and their nature in PI and philosophical offerings were weak.19 In China, the majority examine the targeted research object, i.e., information and of researchers are either researchers of library studies/ communication technologies and ethics. The examination library and information science or librarians/information leads to the fulfillment of hope in the second generation researchers. The information scientists, , of IE. The following six chapters concentrate on studying philosophers, etc., comprising the contemporary western the ethical impacts of information, Internet, and computer IE research team are seriously lacking. This is clearly due to technologies upon a society. Floridi’s information ethics the division of scholarly studies in China and the sporadic focuses on certain concepts, for instance, external and Chinese IE studies as well. semantical views about information, the intrinsic value of the infosphere, the object-oriented programming On the contrary, Floridi embarked upon his academic methodology, and constructionist ethics. Those concepts journey firstly as a philosopher. He then looked into are associated with the basic ethical issues resulting computers from the perspective of information ethics from diversified information technologies, and are and eventually constructed a philosophical foundation appropriately extended here for applications. For example, of information theories. Next, he thoroughly and broadly Floridi proposes a new class of hybrid evil, the “artificial built a well-developed theory on the second-generation evil,” which can complement the traditional distinction information ethics. In his book, he proposed numerous between moral evil and natural evil. Human beings may pioneering viewpoints, which put him in the forefront of the act as agents of natural evils, such as unaware and healthy field. And those views have great implications for Chinese carriers of a contagious disease, and the allegedly natural IE studies. Particularly, many of Floridi’s books and articles, occurrence of disasters such as earthquake, tsunami, for example, his forceful articles advocating for philosophy drought, etc., may result from human blameworthy of information and his Philosophy of Information, are widely negligence or undue interventions to the environment. known in the Chinese academia and have fueled the Furthermore, he introduces a productive initial approach philosophy of information studies in China. The publication which helps to understand personal identity construction and circulation of this book in China will inarguably advance in onlife experience and then proposes an expectation for the scholarship in information ethics. a new ecology of self which completely accommodates the requests of an unspoiled being inhabited in an 5. COMPARISON OF “SELF” UPON WHICH THE infosphere. Then the book examined informational privacy BOOK ELABORATES WITH “SELF-RESTRAINING in the aspects of the ontological interpretation, distributed morality, information , global information IN PRIVACY” IN CHINESE CULTURE. ethics, etc. In principle, this is a serious deliberation of the Given our cultural background, we would like to share our values people hold in an information era. thoughts on Floridi’s interpretations of self from a cross- cultural point of view. Floridi claimed that the IE studies All in all, the book is structured in such a way that the he constructed were in parallel with numerous ethical framework and approaches are complementary and traditions, which is undoubtedly true. In contemporary accentuated, and the book and its chapters are logically China, whether the revival of Confucian studies could organized. This demonstrates the author’s profound lead to moral and ethical reconstruction adaptable to an thinking both in breadth and depth. information society is still a pending issue. It’s generally thought that a liberal information society is prone to collapse 4. THE BOOK WILL HAVE GREAT IMPLICATIONS and slide into chaos while the Confucian model might be FOR THE DEVELOPMENT OF INFORMATION rigidified and eventually suffocated to death. However, the reality is that much wisdom in the Confucian thoughts and ETHICS IN CONTEMPORARY CHINA. other ancient Chinese thoughts is still inspiring in modern The current IE studies in the west have been groundbreaking times. in ethical implications of computer, Internet, and information technologies, a big step further from the Floridi applied “the logic of realization” into developing earlier computer ethics studies. Impressive achievements the three membranes models (corporeal, cognitive, and have been made in different ways. This book is one of the conscious). He thought that it was the self who talked innovative works. However, information ethics is still an about a self and meanwhile realized information becoming emerging cross-discipline in China. Only a few universities self-conscious through selves only. A self is an ultimate offer this course. Chinese researchers mainly focus their technology of negative entropy. Thus, information source studies on computer ethics. In other words, related studies of a self temporarily overcomes the inherent entropy and are concentrated upon prevalent and desirable topics. turns into consciousness and eventually has the ability They find it difficult to tackle the challenging topics, for to narrate stories of a self that emerged while detaching the lack of theoretical and methodological support for gradually from an external reality. Only the mind could philosophy, not to mention studying in an interconnected explain those information structures of a thing, an organic fashion. Those studies simply look into ethical phenomena entity, or a self. This is surprisingly similar to the great and problems created by information and communication thoughts upheld by Chinese philosophical ideas, such as technologies. Clearly, they lack in breadth and depth and “put your heart in your body” (from the Buddhism classic are therefore not counted as legitimate IE studies. Actually, Vajracchedika-sutra) and the Daoist saying, “the nature

SPRING 2015 | VOLUME 14 | NUMBER 2 PAGE 35 APA NEWSLETTER | PHILOSOPHY AND COMPUTERS

lives with me in symbiosis, and everything is with me as 12. Ibid., 389. a whole.” (Zhuangzi ‘Equalizing All Things’) And this is the 13. Adriaans, “Information.” niche that the mind occupies in the universe. 14. Terrell Ward Bynum, “Philosophy in the Information Age,” 175. Admittedly speaking, the two ethics are both similar and 15. Floridi, Ethics of Information, 101. different. China boasts a five-thousand-year-old civilization, 16. Bill Uzgalis, “Information Informs the Field: A Conversation with and the ethical traditions in Confucianism, Daoism, and Luciano Floridi.” Chinese Buddhism are rooted in the Chinese culture. The 17. Floridi, The Philosophy of Information. ancient Chinese paid great attention to the moral function 18. Luciano Floridi, “What Is the Philosophy of Information?” of “self-restraining in privacy” and even regarded it as “the 19. Simon Rogerson and Terrell Ward Bynum, “Information Ethics: way of learning to be moral.” “Self-restraining in privacy” The Second Generation, The Future of Information Systems.” is from The Doctrine of the Mean (Zhong Yong): nothing is 20. The junzi is the ethical exemplar with the virtues making it more visible than the obscure, nothing is plainer than the possible to follow the dao—the way human beings ought to live subtle. Hence, the junzi20 is cautious when he is alone. It their lives. Quoted from David Wong, “Chinese Ethics,” Stanford means that while a person is living or meditating alone, his Encyclopedia of Philosophy, http://plato.stanford.edu/entries/ behaviors should be prudent and moral even though they ethics-chinese. might not be seen. However, in an era when “subjectivity” 21. Floridi, Ethics of Information, 122. is dramatically encroached, is this still possible in reality? BIBLIOGRAPHY Moreover, the early Daoist ethical idea of “inherited Bynum, T. W. “Philosophy in the Information Age.” In Putting Information burden” seems to hear a distant echo in Floridi’s axiological First: Luciano Floridi and the Philosophy of Information, edited by Patrick Allo, 171–93. Wiley-Blackwell, 2010. ecumenism.21 Floridi’s IE presents ethics beyond the center of biological beings. Infosphere-based, it attempts Capurro, Rafael. “Ethical Challenges of the Information Society in the 21st Century.” International Information & Library Review 32 (2000): to center around all beings and see them as inforgs, be 257–76. they living or non-living beings. As a result, it expands the Floridi, Luciano. “What Is the Philosophy of Information?” scope of subjects of value, breaks the anthropocentric and Metaphilosophy 33, no. 1/2 (2002): 123–45. agent-metaphysical grounds, and constructs an ontological Floridi, Luciano. “Information Ethics, Its Nature and Scope.” Computers commitment into moral conducts while we and each and Society 35, no. 2 (2005): 1–3. individual evolving with information technologies, as being Floridi, Luciano. The Philosophy of Information. Oxford: Oxford in the world, stay and meditate alone. That is, even though University Press, 2011. there are no people around, many subjects of value do Floridi, Luciano. The Ethics of Information. Oxford: Oxford University exist. Press, 2013. Floridi, Luciano (ed.), The Onlife Manifesto: Being Human in a NOTES Hyperconnected Era. Springer Open, 2015. 1. Luciano Floridi, The Onlife Manifesto, 2. Floridi, Luciano, and J. W. Sanders. “Mapping the Foundationalist Debate.” In Readings in , 2nd ed., edited by R. Spinello and 2. Luciano Floridi, The Ethics of Information. H. Tavani. Boston, MA: Jones and Bartlett, 2004. 3. Simon Rogerson and Terrell Ward Bynum, “Information Ethics.” Froehlich, Thomas J. “Rafael Capurro and the Challenge of Information 4. Floridi, Ethics of Information, 64. Ethics.” Intl. Inform. & Libr. Rev. 32 (2000): 277–82. Rogerson, S., and T. W. Bynum. “Information Ethics: The Second 5. Thomas J. Froehlich, “Rafael Capurro and the Challenge of Generation, the Future of Information Systems.” UK Academy for Information Ethics,” 279. Information Systems Conference, 1996. http://www.cms.dmu.ac.uk/ 6. Floridi, Ethics of Information, 19. resources/general/discipline/ie_sec_ gen.html. 2015-01-26. 7. Ibid., 65. Shannon, Claude E. “A Mathematical Theory of Communication.” Bell System Technical Journal 27 (1948): 379–423, 623–56. 8. Ibid., 66. Uzgalis, Bill. “Information Informs the Field: A Conversation with 9. Ibid., 67. Luciano Floridi.” APA Newsletter on Philosophy and Computers 2, no. 1 (Fall 2002): 72–77. 10. Pieter Adriaans, “Information,” Stanford Encyclopedia of Philosophy. Wong, David. “Chinese Ethics.” Stanford Encyclopedia of Philosophy, February 2, 2015. http://plato.stanford.edu/entries/ethics-chinese. 11. Claude E. Shannon, “A Mathematical Theory of Communication,” 390.

PAGE 36 SPRING 2015 | VOLUME 14 | NUMBER 2