DOCSLIB.ORG

Facial Recognition Is Improving — but the Bigger That Had Been Scanned

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Facial Recognition Is Improving — but the Bigger That Had Been Scanned Feature KELVIN CHAN/AP/SHUTTERSTOCK KELVIN The Metropolitan Police in London used facial-recognition cameras to scan for wanted people in February. Fussey and Murray listed a number of ethical and privacy concerns with the dragnet, and questioned whether it was legal at all. And they queried the accuracy of the system, which is sold by Tokyo-based technology giant NEC. The software flagged 42 people over 6 trials BEATING that the researchers analysed; officers dis- missed 16 matches as ‘non-credible’ but rushed out to stop the others. They lost 4 people in the crowd, but still stopped 22: only 8 turned out to be correct matches. The police saw the issue differently. They BIOMETRIC BIAS said the system’s number of false positives was tiny, considering the many thousands of faces Facial recognition is improving — but the bigger that had been scanned. (They didn’t reply to issue is how it’s used. By Davide Castelvecchi Nature’s requests for comment for this article.) The accuracy of facial recognition has improved drastically since ‘deep learning’ tech- hen London’s Metropolitan extracted key features and compared them niques were introduced into the field about a Police tested real-time to those of suspects from a watch list. “If there decade ago. But whether that means it’s good facial-recognition technology is a match, it pulls an image from the live feed, enough to be used on lower-quality, ‘in the wild’ between 2016 and 2019, they together with the image from the watch list.” images is a hugely controversial issue. And invited Daragh Murray to moni- Officers then reviewed the match and decided questions remain about how to transparently tor some of the trials from a con- whether to rush out to stop the ‘suspect’ and, evaluate facial-recognition systems. trol room inside a police van. occasionally, arrest them. In 2018, a seminal paper by computer scien- “It’s like you see in the Scotland Yard, as the headquarters of the tists Timnit Gebru, then at Microsoft Research Wmovies,” says Murray, a legal scholar at the Uni- London police force is sometimes known, had in New York City and now at Google in Moun- versity of Essex in Colchester, UK. As cameras commissioned Murray and his University of tain View, California, and Joy Buolamwini at the scanned passers-by in shopping centres or Essex colleague Pete Fussey, a sociologist, to Massachusetts Institute of Technology in Cam- public squares, they fed images to a computer conduct an independent study of its dragnet. bridge found that leading facial-recognition inside the van. Murray and police officers saw But their results1, published in July 2019, might software packages performed much worse at the software draw rectangles around faces not have been quite what the law-enforcement identifying the gender of women and people as it identified them in the live feed. It then agency had hoped for. of colour than at classifying male, white faces2. Nature | Vol 587 | 19 November 2020 | 347 ©2020 Spri nger Nature Li mited. All ri ghts reserved. Feature Concerns over demographic bias have since passport control at national borders. small group of vendors where false positives been quoted frequently in calls for morato- One measure of progress is the Face Recogni- based on demographic differentials were unde- riums or bans of facial-recognition software. tion Vendor Test, an independent benchmark- tectable”, but that match rates could be compro- In June, the world’s largest scientific com- ing assessment that the US National Institute of mised by outdoor, poorly lit or grainy images. puting society, the Association for Com- Standards and Technology (NIST) in Gaithers- puting Machinery in New York City, urged a burg, Maryland, has been conducting for two False faces suspension of private and government use decades. Dozens of laboratories, both com- One-to-one verification, such as recognizing of facial-recognition technology, because mercial and academic, have voluntarily taken the rightful owner of a passport or smart- of “clear bias based on ethnic, racial, gender, part in the latest round of testing, which began phone, has become extremely accurate; here, and other human characteristics”, which it in 2018 and is ongoing. NIST measures the per- artificial intelligence is as skilful as the sharp- said injured the rights of individuals in spe- formance of each lab’s software package on its est-eyed humans. In this field, cutting-edge cific demographic groups. Axon, a maker of own image data sets, which include frontal and research focuses on detecting malevolent body cameras worn by police officers across profile police mugshots, and pictures scraped attacks. The first facial-recognition systems the United States, has said that facial recogni- from the Internet. (The US technology giants for unlocking phones, for example, were easily tion isn’t accurate enough to be deployed in Amazon, Apple, Google and Facebook have fooled by showing the phone a photo of the its products. Some US cities have banned the not taken part in the test.) owner, Jain says; 3D face recognition does bet- use of the technology in policing, and US law- In reports released late last year, the NIST ter. “Now the biggest challenge is very-high- makers have proposed a federal moratorium. team described massive steps forward in the quality face masks.” In one project, Jain and his Companies say they’re working to fix the technology’s performance during 2018, both collaborators are working on detecting such biases in their facial-recognition systems, and for one-to-many searches3 and for one-to-one impersonators by looking for skin texture. some are claiming success. But many research- verification4 (see also go.nature.com/35pku9q). But one-to-many verification, as Murray ers and activists are deeply sceptical. They argue “We have seen a significant improvement in found, isn’t so simple. With a large enough that even if the technology surpasses some face-recognition accuracy,” says Craig Watson, watch list, the number of false positives benchmark in accuracy, that won’t assuage an electrical engineer who leads NIST’s image flagged up can easily outweigh the true hits. deeper concerns that facial-recognition tools group. “We know that’s largely because of con- This is a problem when police must make are used in discriminatory ways. volutional neural networks,” he adds, a type of quick decisions about stopping someone. But deep neural network that is especially efficient mistakes also occur in slower investigations. More accurate but still biased at recognizing images. In January, Robert Williams was arrested at his Facial-recognition systems are often propri- The best algorithms can now identify peo- house in Farmington Hills, Michigan, after a etary and swathed in secrecy, but specialists ple from a profile image taken in the wild police facial-recognition system misidenti- say that most involve a multi-stage process — matching it with a frontal view from the fied him as a watch thief on the basis of blurry (see ‘How facial recognition works’) using deep database — about as accurately as the best surveillance footage of a Black man, which learning to train massive neural networks on it matched to his driving licence. The Amer- large sets of data to recognize patterns. “Every- “Systems are being brought ican Civil Liberties Union (ACLU), a non-profit body who does face recognition now uses deep organization in New York City, filed a com- learning,” says Anil Jain, a computer scientist to the wild without a plaint about the incident to Detroit police in at Michigan State University in East Lansing. proper evaluation of their June, and produced a video in which Williams The first stage in a typical system locates performance.” recounts what happened when a detective one or more faces in an image. Faces in the feed showed him the surveillance photos on paper. from a surveillance camera might be viewed in a “I picked that paper up, held it next to my face range of lighting conditions and from different facial-recognition software from a decade and said, ‘This is not me. I hope y’all don’t think angles, making them harder to recognize than ago could recognize frontal images, NIST all Black people look alike.’ And then he said: in a standard passport photo, for instance. The found. Recognizing a face in profile “has been ‘The computer says it’s you,’” Williams said. He algorithm will have been trained on millions a long-sought milestone in face recognition was released after being detained for 30 hours. of photos to locate ‘landmarks’ on a face, such research”, the NIST researchers wrote. ACLU attorney Phil Mayor says the technology as the eyes, nose and mouth, and it distils the But NIST also confirmed what Buolamwini should be banned. “It doesn’t work, and even information into a compact file, ranging from and Gebru’s gender-classification work sug- when it does work, it remains too dangerous less than 100 bytes to a few kilobytes in size. gested: most packages tended to be more a tool for governments to use to surveil their The next task is to ‘normalize’ the face, artifi- accurate for white, male faces than for peo- own citizens for no compelling return,” he says. cially rotating it into a frontal, well-illuminated ple of colour or for women5. In particular, Shortly after the ACLU complaint, Detroit view. This produces a set of facial ‘features’ faces classified in NIST’s database as African police chief James Craig acknowledged that the that can be compared with those extracted American or Asian were 10–100 times more software, if used by itself, would misidentify from an existing database of faces.
Load more

Recommended publications
  • When Diving Deep Into Operationalizing the Ethical Development of Artificial Intelligence, One Immediately Runs Into the “Fractal Problem”
    When diving deep into operationalizing the ethical development of artificial intelligence, one immediately runs into the “fractal problem”. This may also be called the “infinite onion” problem.1 That is, each problem within development that you pinpoint expands into a vast universe of new complex problems. It can be hard to make any measurable progress as you run in circles among different competing issues, but one of the paths forward is to pause at a specific point and detail what you see there. Here I list some of the complex points that I see at play in the firing of Dr. Timnit Gebru, and why it will remain forever after a really, really, really terrible decision. The Punchline. The firing of Dr. Timnit Gebru is not okay, and the way it was done is not okay. It appears to stem from the same lack of foresight that is at the core of modern technology,2 and so itself serves as an example of the problem. The firing seems to have been fueled by the same underpinnings of racism and sexism that our AI systems, when in the wrong hands, tend to soak up. How Dr. Gebru was fired is not okay, what was said about it is not okay, and the environment leading up to it was -- and is -- not okay. Every moment where Jeff Dean and Megan Kacholia do not take responsibility for their actions is another moment where the company as a whole stands by silently as if to intentionally send the horrifying message that Dr. Gebru deserves to be treated this way.
    [Show full text]
  • Bias and Fairness in NLP
    Bias and Fairness in NLP Margaret Mitchell Kai-Wei Chang Vicente Ordóñez Román Google Brain UCLA University of Virginia Vinodkumar Prabhakaran Google Brain Tutorial Outline ● Part 1: Cognitive Biases / Data Biases / Bias laundering ● Part 2: Bias in NLP and Mitigation Approaches ● Part 3: Building Fair and Robust Representations for Vision and Language ● Part 4: Conclusion and Discussion “Bias Laundering” Cognitive Biases, Data Biases, and ML Vinodkumar Prabhakaran Margaret Mitchell Google Brain Google Brain Andrew Emily Simone Parker Lucy Ben Elena Deb Timnit Gebru Zaldivar Denton Wu Barnes Vasserman Hutchinson Spitzer Raji Adrian Brian Dirk Josh Alex Blake Hee Jung Hartwig Blaise Benton Zhang Hovy Lovejoy Beutel Lemoine Ryu Adam Agüera y Arcas What’s in this tutorial ● Motivation for Fairness research in NLP ● How and why NLP models may be unfair ● Various types of NLP fairness issues and mitigation approaches ● What can/should we do? What’s NOT in this tutorial ● Definitive answers to fairness/ethical questions ● Prescriptive solutions to fix ML/NLP (un)fairness What do you see? What do you see? ● Bananas What do you see? ● Bananas ● Stickers What do you see? ● Bananas ● Stickers ● Dole Bananas What do you see? ● Bananas ● Stickers ● Dole Bananas ● Bananas at a store What do you see? ● Bananas ● Stickers ● Dole Bananas ● Bananas at a store ● Bananas on shelves What do you see? ● Bananas ● Stickers ● Dole Bananas ● Bananas at a store ● Bananas on shelves ● Bunches of bananas What do you see? ● Bananas ● Stickers ● Dole Bananas ● Bananas
    [Show full text]
  • Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification∗
    Proceedings of Machine Learning Research 81:1{15, 2018 Conference on Fairness, Accountability, and Transparency Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification∗ Joy Buolamwini [email protected] MIT Media Lab 75 Amherst St. Cambridge, MA 02139 Timnit Gebru [email protected] Microsoft Research 641 Avenue of the Americas, New York, NY 10011 Editors: Sorelle A. Friedler and Christo Wilson Abstract who is hired, fired, granted a loan, or how long Recent studies demonstrate that machine an individual spends in prison, decisions that learning algorithms can discriminate based have traditionally been performed by humans are on classes like race and gender. In this rapidly made by algorithms (O'Neil, 2017; Citron work, we present an approach to evaluate and Pasquale, 2014). Even AI-based technologies bias present in automated facial analysis al- that are not specifically trained to perform high- gorithms and datasets with respect to phe- stakes tasks (such as determining how long some- notypic subgroups. Using the dermatolo- one spends in prison) can be used in a pipeline gist approved Fitzpatrick Skin Type clas- that performs such tasks. For example, while sification system, we characterize the gen- face recognition software by itself should not be der and skin type distribution of two facial analysis benchmarks, IJB-A and Adience. trained to determine the fate of an individual in We find that these datasets are overwhelm- the criminal justice system, it is very likely that ingly composed of lighter-skinned subjects such software is used to identify suspects. Thus, (79:6% for IJB-A and 86:2% for Adience) an error in the output of a face recognition algo- and introduce a new facial analysis dataset rithm used as input for other tasks can have se- which is balanced by gender and skin type.
    [Show full text]
  • AI Principles 2020 Progress Update
    AI Principles 2020 Progress update AI Principles 2020 Progress update Table of contents Overview ........................................................................................................................................... 2 Culture, education, and participation ....................................................................4 Technical progress ................................................................................................................... 5 Internal processes ..................................................................................................................... 8 Community outreach and exchange .....................................................................12 Conclusion .....................................................................................................................................17 Appendix: Research publications and tools ....................................................18 Endnotes .........................................................................................................................................20 1 AI Principles 2020 Progress update Overview Google’s AI Principles were published in June 2018 as a charter to guide how we develop AI responsibly and the types of applications we will pursue. This report highlights recent progress in AI Principles implementation across Google, including technical tools, educational programs and governance processes. Of particular note in 2020, the AI Principles have supported our ongoing work to address
    [Show full text]
  • Distributed Blackness Published Just Last Year (2020, New York University Press), Dr
    Transcript The Critical Technology Episode 4: Distributed Blackness Published just last year (2020, New York University Press), Dr. Andre Brock Jr’s Distributed Blackness: African American Cybercultures has already become a highly influential, widely cited, must read text for anyone interested in digital culture and technology. Through a theoretically rich and methodologically innovative analysis, Brock positions Blackness at the centre of Internet technoculture, its technological designs and infrastructures, its political economic underpinnings, its cultural significance and its emergent informational identities. Sara Grimes 0:00 Like many of you, as the fall 2020 semester was wrapping up, my Twitter feed was suddenly flooded with the news that Dr. Timnit Gebru had been fired from Google. A world renowned computer scientist, leading AI ethics researcher, and co-founder of Black in AI. Dr. Gebru had recently completed research raising numerous ethical concerns with the type of large scale AI language models used in many Google products. Her expulsion from Google highlights the discrimination and precarity faced by black people, especially black women, working in the tech industries and in society at large. On the one hand, Dr. Gebru’s treatment by Google execs made visible through screenshots and conversations posted on Twitter was met with an outpouring of public support from fellow Google employees, tech workers, academics and concerned citizens. The #ISupportTimnit was often paired with #BelieveBlackWomen, connecting this event with the systematic anti-black racism and sexism found across technoculture. On the other hand, since December, Dr. Gebru and her supporters have also been targeted by Twitter trolls, and subjected to a bombardment of racist and sexist vitriol online.
    [Show full text]
  • Saving Face: Investigating the Ethical Concerns of Facial Recognition Auditing
    Saving Face: Investigating the Ethical Concerns of Facial Recognition Auditing Inioluwa Deborah Raji Timnit Gebru Margaret Mitchell University of Toronto Google Research Google Research Canada United States United States [email protected] [email protected] [email protected] Joy Buolamwini Joonseok Lee Emily Denton MIT Media Lab. Google Research Google Research United States United States United States [email protected] [email protected] [email protected] ABSTRACT the positive uses of this technology includes claims of “understand- Although essential to revealing biased performance, well inten- ing users”, “monitoring or detecting human activity”, “indexing and tioned attempts at algorithmic auditing can have effects that may searching digital image libraries”, and verifying and identifying harm the very populations these measures are meant to protect. subjects “in security scenarios” [1, 8, 16, 33]. This concern is even more salient while auditing biometric sys- The reality of FPT deployments, however, reveals that many tems such as facial recognition, where the data is sensitive and of these uses are quite vulnerable to abuse, especially when used the technology is often used in ethically questionable manners. for surveillance and coupled with predatory data collection prac- We demonstrate a set of five ethical concerns in the particular case tices that, intentionally or unintentionally, discriminate against of auditing commercial facial processing technology, highlighting marginalized groups [2]. The stakes are particularly high when additional design considerations and ethical tensions the auditor we consider companies like Amazon and HireVue, who are selling needs to be aware of so as not exacerbate or complement the harms their services to police departments or using the technology to help propagated by the audited system.
    [Show full text]
  • The Privatization of AI Research(-Ers): Causes and Potential Consequences
    The Privatization of AI Research(-ers): Causes and Potential Consequences – From university-industry interaction to public research brain-drain? – „ „ Roman Jurowetzki , Daniel S. Hain , Juan Mateos-Garcia†, and Konstantinos Stathoulopoulos† „ Aalborg University Business School, DK †Nesta, UK February 15, 2021 Abstract: The private sector is playing an increasingly important role in basic Ar- tificial Intelligence (AI) R&D. This phenomenon, which is reflected in the perception of a brain drain of researchers from academia to industry, is raising concerns about a privatisation of AI research which could constrain its societal benefits. We contribute to the evidence base by quantifying transition flows between industry and academia and studying its drivers and potential consequences. We find a growing net flow of researchers from academia to industry, particularly from elite institutions into technology companies such as Google, Microsoft and Facebook. Our survival regression analysis reveals that researchers working in the field of deep learning as well as those with higher average impact are more likely to transition into industry. A difference-in-differences analysis of the effect of switching into industry on a researcher’s influence proxied by citations indicates that an initial increase in impact declines as researchers spend more time in industry. This points at a privatisation of AI knowledge compared to a counterfactual where those high-impact researchers had remained in academia. Our findings highlight the importance of strengthening the public AI research sphere in order to ensure that the future of this powerful technology is not dominated by private interests. Keywords: AI, university-industry interaction, researcher careers, private research, bibliometrics 1 1 Introduction “The regulatory environment for technology is often led by the people who control the technology” — Zoubin Ghahramani In December 2020, renowned AI researcher Timnit Gebru was dismissed from her po- sition as Ethics Co-Lead in Google Brain, Google’s AI research unit (Hao, 2020b).
    [Show full text]
  • Nist-Ai-Rfi-Microsoft-001.Pdf | NIST
    Microsoft’s Response to the Request for Information of the National Institute of Standards and Technology Developing Artificial Intelligence Standards I. Introduction and Executive Summary Microsoft Corporation (“Microsoft”) welcomes the opportunity to provide comments to the National Institute of Standards and Technology (“NIST”) in response to its May 1, 2019, request for information (“RFI”).1 The RFI seeks comments on the current state, plans, challenges, and opportunities regarding the development and availability of artificial intelligence (“AI”) technical standards and related tools, as well as priority areas for federal involvement in AI standards‐related activities. Microsoft supports NIST’s efforts to meet its obligations under the February 11, 2019, Executive Order on Maintaining American Leadership in Artificial Intelligence (“Executive Order on AI”), including driving development of appropriate technical standards that “enable the creation of new AI‐related industries and the adoption of AI by today’s industries.”2 AI has the potential to transform economies and to address societal challenges by enhancing human decision‐making processes with additional insights and intelligence, leading to better outcomes and improvements in every aspect of people’s lives. Like many emerging technologies, however, AI can be misused in ways that can create individual, organizational, and societal concerns. As a leading supplier of AI solutions, Microsoft is committed to responsible AI innovation. We recognize the need for sustained, constructive engagement among industry, government, academia, and other stakeholders in developing and using AI. We recommend policymakers in the United States to commit to high‐level ethical and moral principles necessary to support the development and use of AI in a manner that maximizes the benefits of the technology while protecting individuals and society from its potential risks.
    [Show full text]
  • Bias and Representation in Artificial Intelligence
    CIBER Focus Interview Series Video Annotation Bias and Representation in Artificial Intelligence Link: https://www.youtube.com/watch?v=e2i1iADw7A0 Length: 18:17 Production Date/Year: December 5, 2018 Keyword Topics: Artificial Intelligence, Representation, Bias, STEM, Discrimination, Guest Info: Dr. Timnit Gebru Dr. Timnit Gebru recently began working as a Research Scientist at Google AI after completing her post-doc research in the Fairness Accountability Transparency and Ethics group at Microsoft Research. Her research currently targets ethical considerations of data mining projects and methods to audit and mitigate bias. Dr. Gebru received her Ph.D. from the Stanford Artificial Intelligence Laboratory and her work has been featured in the New York Times, Economist, and MIT Tech Review. She is the co-founder of Black in AI, where she works to both increase diversity in the artificial intelligence field and reduce the negative impacts of racial bias in data. Video Summary/Synopsis: 1:53 - Gebru displays her findings of racial bias being present in facial recognition software/artificial intelligence and what greater effects this may have on society. 4:20 - Gebru hypothesizes the reasons why these discrepancies in artificial intelligence occur. 5:42 - Gebru discusses whether there are other areas of A.I. outside of facial recognition software that are being affected by inherent racial bias. 7:10 - Gebru outlines what can be done going forward to reduce the issue of racial bias in artificial intelligence software. 10:12 - Gebru explains what her Black in AI organization does and how it started. 14:32 - Gebru explains the effects that a lack of representation has had on the STEM field while going on to explain what the business community can do to help make change.
    [Show full text]
  • Strategies for Collecting Sociocultural Data in Machine Learning
    Lessons from Archives: Strategies for Collecting Sociocultural Data in Machine Learning Eun Seo Jo Timnit Gebru Stanford University Google [email protected] [email protected] ABSTRACT A growing body of work shows that many problems in fairness, accountability, transparency, and ethics in machine learning sys- tems are rooted in decisions surrounding the data collection and annotation process. In spite of its fundamental nature however, data collection remains an overlooked part of the machine learning (ML) pipeline. In this paper, we argue that a new specialization should be formed within ML that is focused on methodologies for data collection and annotation: efforts that require institutional frame- works and procedures. Specifically for sociocultural data, parallels can be drawn from archives and libraries. Archives are the longest standing communal effort to gather human information and archive scholars have already developed the language and procedures to address and discuss many challenges pertaining to data collection such as consent, power, inclusivity, transparency, and ethics & pri- vacy. We discuss these five key approaches in document collection practices in archives that can inform data collection in sociocultural ML. By showing data collection practices from another field, we encourage ML research to be more cognizant and systematic in data collection and draw from interdisciplinary expertise. CCS CONCEPTS • Computing methodologies → Machine learning. KEYWORDS datasets, machine learning, ML fairness, data collection, sociocul- tural data, archives ACM Reference Format: Figure 1: Article from LIFE magazine (Dec. 1941) with two Eun Seo Jo and Timnit Gebru. 2020. Lessons from Archives: Strategies images advising identifiable phenotype differences between for Collecting Sociocultural Data in Machine Learning.
    [Show full text]
  • The Role of Workers in AI Ethics and Governance
    The Role of Workers in AI Ethics and Governance By Nataliya Nedzhvetskaya and JS Tan Abstract (250 words) While the role of states, corporations, and international organizations in AI governance has been extensively theorized, the role of workers has received comparatively little attention. This chapter looks at the role that workers play in identifying and mitigating harms from AI technologies. Harms are the causally assessed “impacts” of technologies. They arise despite technical reliability and are not a result of technical negligence but rather of normative uncertainty around questions of safety and fairness in complex social systems. There is high consensus in the AI ethics community on the benefits of reducing harms but less consensus on mechanisms for determining or addressing harms. This lack of consensus has resulted in a number of collective actions by workers protesting how harms are identified and addressed in their workplace. We theorize the role of workers within AI governance and construct a model of harm reporting processes in AI workplaces. The harm reporting process involves three steps: identification, the governance decision, and the response. Workers draw upon three types of claims to argue for jurisdiction over questions of AI governance: subjection, control over the product of one’s labor, and proximate knowledge of systems. Examining the past decade of AI-related worker activism allows us to understand how different types of workers are positioned within a workplace that produces AI systems, how their position informs their claims, and the place of collective action in staking their claims. This chapter argues that workers occupy a unique role in identifying and mitigating harms caused by AI systems.
    [Show full text]
  • Image Cropping on Twitter: Fairness Metrics, Their Limitations, and the Importance of Representation, Design, and Agency
    Image Cropping on Twitter: Fairness Metrics, their Limitations, and the Importance of Representation, Design, and Agency KYRA YEE∗, UTHAIPON TANTIPONGPIPAT∗, and SHUBHANSHU MISHRA∗, Twitter, USA Twitter uses machine learning to crop images, where crops are centered around the part predicted to be the most salient. In fall 2020, Twitter users raised concerns that the automated image cropping system on Twitter favored light-skinned over dark-skinned individuals, as well as concerns that the system favored cropping woman’s bodies instead of their heads. In order to address these concerns, we conduct an extensive analysis using formalized group fairness metrics. We find systematic disparities in cropping and identify contributing factors, including the fact that the cropping based on the single most salient point can amplify the disparities because of an effect we term argmax bias. However, we demonstrate that formalized fairness metrics and quantitative analysis on their own are insufficient for capturing the risk of representational harm in automatic cropping. We suggest the removalof saliency-based cropping in favor of a solution that better preserves user agency. For developing a new solution that sufficiently address concerns related to representational harm, our critique motivates a combination of quantitative and qualitative methods that include human-centered design. CCS Concepts: • General and reference ! Experimentation; • Human-centered computing ! HCI theory, concepts and models; • Applied computing ! Sociology; • Computing methodologies ! Computer vision; • Social and professional topics ! User characteristics. Additional Key Words and Phrases: image cropping, demographic parity, representational harm, ethical HCI, fairness in machine learning 1 INTRODUCTION Automated image cropping (or smart image cropping) [1] is a task to, given a viewport dimension or aspect ratio and an image, crop the image such that the image fits the viewport or aspect ratiowidth/height ( ) while ensuring that its most relevant or interesting parts are within the viewport.
    [Show full text]