Artificial Intelligence for Students in Postsecondary Education
Total Page:16
File Type:pdf, Size:1020Kb
AI MATTERS, VOLUME 6, ISSUE 3 DECEMBER 2020 Artificial Intelligence for Students in Postsecondary Education: A World of Opportunity Natalina Martiniello (University of Montreal´ (School of Optometry), Centre for Interdisci- plinary Research in Rehabilitation of Greater Montreal´ (CRIR), Adaptech Research Network; [email protected]) Jennison Asuncion (Adaptech Research Network; [email protected]) Catherine Fichten (Adaptech Research Network, Dawson College, McGill University (De- partment of Psychiatry), Jewish General Hospital (Department of Psychiatry, Behavioral Psy- chotherapy and Research Unit); catherine.fi[email protected]) Mary Jorgensen (Adaptech Research Network; [email protected]) Alice Havel (Adaptech Research Network, Dawson College; [email protected]) Maegan Harvison (Adaptech Research Network; [email protected]) Anick Legault (Adaptech Research Network, Dawson College; acle- [email protected]) Alex Lussier (Adaptech Research Network; [email protected]) Christine Vo (Adaptech Research Network; [email protected]) DOI: 10.1145/3446243.3446250 Abstract unique needs of users. While the practical and mundane benefits of artificial intelligence sys- AI-based apps can facilitate learning for all tems are taken for granted by society at large post-secondary students and may also be (e.g. speech recognition, real-time caption- useful for students with disabilities. Here we ing (Mathur, Gill, Yadav, Mishra, & Bansode, share some reflections from discussions that 2017), linguistic translation (Kapoor, 2020), or took place during two advisory board meet- organizing tools such as calendars and “to ings on the use of such apps for students with do” lists (Canbek & Mutlu, 2016)), often over- disabilities at the post-secondary level. looked are the potential benefits to specific Keywords: college and university students populations who would otherwise be depen- with disabilities, artificial intelligence apps, dent upon third parties for support. For exam- mobile AI apps ple, artificial intelligence has been success- fully harnessed to provide people with men- tal health concerns an always-available, point- Introduction in-time responsive, supplementary or interme- Intelligent technologies (such as smartphones diate support system (Inkster, Sarda, & Sub- and tablets which incorporate principles of uni- ramanian, 2018). The available evidence at versal design) have the potential to increase present indicates that individuals with disabili- the inclusion of students with disabilities and ties are conspicuously excluded from AI train- other diverse learners in different aspects of ing models and have not yet become mean- post-secondary education. Artificial intelli- ingful contributors to, or beneficiaries of, the gence (“AI”), which for our purposes includes ongoing discussions surrounding artificial in- “computing systems that are able to engage telligence and machine learning (Lillywhite & in human-like processes such as learning, Wolbring, 2020). adapting, synthesizing, self-correction and use of data for complex processing tasks” Despite advances in inclusive education prac- (Popenici & Kerr, 2017), is in many respects tices, the reality is that for many of the 10%- the bedrock of more universally accessible en- 20% (Eagan et al., 2017; Fichten et al., 2018; gagement, because it has the potential to per- Snyder, De Brey, & Dillow, 2016) of students mit technology to adapt to the diversity and with disabilities, barriers to accessing post- secondary education continue to persist. In- Copyright c 2020 by the author(s). accessible websites, commonplace at some of 17 AI MATTERS, VOLUME 6, ISSUE 3 DECEMBER 2020 the world’s most renowned institutions, con- The Advisory Board Meetings tinue to pose significant accessibility barriers (Huffington, Copeland, Devany, Parker-Gills, We invited stakeholders from within our net- & Patel, 2020). In spite of increased use of work who could provide insights on the use automatic speech recognition, Deaf and hard of artificial intelligence at the post-secondary of hearing students often do not have ade- level from diverse perspectives. In total, this quate access to qualified real-time captioning included 38 individuals: 7 students, 3 disabil- (Butler, Trager, & Behm, 2019). Even where ity/accessibility service providers, 14 faculty captioning or interpreters are available, Deaf members (some with and some without dis- and hard of hearing students may lag be- abilities), 9 technology experts, and 5 tech- hind their peers due to the increased cognitive nology users with disabilities. Two advisory load of processing the visual translation of au- board meetings were held in May 2020, us- dio information, visual attention limits associ- ing the Zoom (San Jose, CA) videoconferenc- ated with trying to track dispersed visual infor- ing system to accommodate for the different mation in the classroom, difficulties engaging time zones in which the international atten- in discussions, and limited social interactions dees resided. Spontaneously, attendees di- with peers, among other factors (Kushalnagar, vided themselves roughly in half into one of 2019). the two sessions. The following overarching questions were distributed in advance of the The initiative we report in this article had its meetings and provided a framework for the genesis in prior research exploring the is- discussion: sue of fairness of AI for people with dis- abilities (in employment, education, public safety, and healthcare (Trewin et al., 2019)). 1. What AI-based smartphone or tablet tech- Building on our previous research regard- nologies have you come across that are cur- ing the accessibility of information and com- rently used by college and university stu- puting technologies for post-secondary stu- dents with disabilities? Which ones work dents (Fichten, Asuncion, & Scapin, 2014; well? Which ones do not? Thomson, Fichten, Havel, Budd, & Asun- cion, 2015) and the increasing use and util- 2. What AI-based smartphone and tablet tech- ity of ubiquitous mobile technologies, such as nologies are out there but are rarely consid- smartphones and tablets in post-secondary ered even though they could help college education (Fichten et al., 2019), the ques- and university students with disabilities do tion arose as to the impact and adoption academic work? of new AI-based technologies. We there- fore convened an advisory group to gather input from post-secondary students, con- 3. What are some functions you can foresee sumers with disabilities, post-secondary dis- in the future that AI-based smartphone and ability/accessibility service providers, faculty, tablet technologies could perform to assist and technology experts to explore how college students with disabilities succeed in higher and university students with diverse disabili- education? (i.e. It would be great if an ties (visual or hearing impairments, learning AI-based smartphone and tablet technology disabilities, attention deficit hyperactivity disor- could help a student do X.) der, etc.) could benefit from AI-based smart- phone and tablet apps, and how these could facilitate student academic success. In sum- Summaries of comments from members of the marizing the themes that emerged during the advisory board, including the identification of advisory board meetings, we hope to lay the specific tools and resources (see Adaptech groundwork for more in-depth and targeted ini- Research Network, 2020), were then pre- tiatives to objectively evaluate the effective- pared. The resulting summaries, grouped and ness of the tools currently being used, and to categorized by Adaptech Research Network explore the feasibility and potential success of team members according to theme and the future innovations within the post-secondary nature of the application, are described more education context. fully below. 18 AI MATTERS, VOLUME 6, ISSUE 3 DECEMBER 2020 Advisory Board Meeting Themes significant disabilities (e.g. those who are functionally blind) may actually develop more General observations sophisticated technological capabilities than Those who need AI the most are also the their peers. For example, Fichten, Asuncion, most likely not to be able to use AI: A Barile, Ferraro, and Wolforth(2009) found that recurring comment was that systems relying students who were functionally blind felt sig- on artificial intelligence have the potential to nificantly more comfortable utilizing technol- open the world to individuals who cannot ac- ogy in the classroom than students who had cess information or environments in the same low vision. Nonetheless, many instances were manner as everyone else. Unfortunately, be- noted whereby students who could benefit cause AI is based on training data and learn- from the use of AI-enabled technologies (such ing analytics derived from a generally able- as Seeing AI) were simply unaware that such bodied population, those who veer from what tools even existed. Learning that the technolo- is viewed as ’the norm’ and who have distinc- gies exist and how to effectively use them is, tive voice, facial, or movement characteristics for many, a critical step in the transition from are likely to be misunderstood (or not under- high school to post-secondary education. The stood at all) by AI capture processes. As one lack of training is especially problematic for individual described it: students who acquire a disability later in life and who do not have previous experience with AI programs are made up of algorithms,