Free beta access. 50 free AI tokens each week. Optional top-ups available. See what is included
Opening Word document…

AI development homogenizes disability and labels disabled bodies as “Burdens”

Shew 20 - (Ashley Shew 20 ; Ashley Shew is an Associate Professor in Science, Technology, and Society at Virginia Tech, where she participates in the STS PhD program, the Medicine & Society minor, the Disability Studies minor, the Bioethics graduate certificate, and Integrative Graduate Education Program on Regenerative Medicine. Published: Mar 1, 2020 ; Ableism, Technoableism, and Future AI. Retrieved from https://par.nsf.gov/biblio/10165545. IEEE Technology and Society Magazine 39.1 Web. doi:10.1109/MTS.2020.2967492.) Dōgen+Barney)

The latest set of technological solutions to the perceived problems of disabled people is Artificial Intelligence and machine learning. But like other technologies discussed in this essay, Artificial Intelligence research and AIs currently under development contain several inscriptions of technoableism that I want to warn against. I also want to gesture to an emerging frame for considering appropriate disability technology, called Crip Technoscience [16]. The current mood is one of optimism about machine learning and Artificial Intelligence for disability applications. Headlines scream about it: “How AI Could Transform the Lives of Disabled People” [10], “How Artificial Intelligence Can Help With Disabilities” [19], and “Artificial Intelligence Poised to Improve Lives of People with Disabilities” [46]. Sometimes these headlines acknowledge AI’s potential for helping people as they age (correctly assuming that older folks are more likely to have disabilities), hyping the potential for robot caregivers and AI emotional support “animals” — AI that can take the place of human carers or friends, often at a lower financial cost: “Artificial Intelligence Could Improve How We Age” [29], “Can China Win the Artificial Intelligence Race by Serving the Elderly?” [18], and “Robot Caregivers are Saving the Elderly from Lives of Loneliness” [43]. These breathless headlines make clear the assumptions that ground the articles that follow them — assumptions about what it is to be disabled and what it is to age. (These articles tend to lump disability and aging together, and I therefore examine the two together, but there are important differences between these demographic groups in terms of stigma, access, convenience, societal narrative, and consideration.) Like earlier technological interventions on disabled lives, these technologies rely on assumptions about the disabled and aging individual rather than the environment. For example, ideas about the importance of independence factor heavily into coverage of AI for disabled and older folks, and are particularly clear in discussions around autonomous vehicles, companion robots (like Paro), home-based applications (like Alexa), and caretaking robots. All of these technologies emphasize aging and disability as problems for individual consumers, rather than problems for society.12 Why should the lack of accessible transport for people who cannot drive be a problem for the non-driver rather than for society? Why should the loneliness and isolation experienced by people who cannot easily leave their homes be solved by in-home robots rather than by better planning and infrastructure within our communities? AI designers are clearly focused on changing the individual to fit the environment rather than the other way around. For example, many of the “conveniences” offered by AI applications are mostly conveniences for the benefit of the user’s family, rather than the users themselves. These technologies are framed as unburdening families from worry by enabling them to remotely monitor or check in on their loved one’s condition; as reducing the financial costs of hiring human caretakers; and as combating the user’s isolation and loneliness, thus requiring fewer visits from family members. When AI technology is framed as helping the family, it positions disabled and old people as recipients of care rather than as agents in their own lives, and it decenters them and their experiences, instead moving the nondisabled family members into the center of the story (which also fails to recognize that disabled people can have disabled family members and can be caregivers themselves).

This same centering of ability is produced by the narratives’ focus on independence as the main goal of disability design. Indeed, the nearly obsessive focus on independence is another “tell” for ableism. As Elliott Kukla writes: We are born needing care and die needing care, and I am no exception. At brief moments in the middle of life, we hold the illusion of independence, but we are always driving on roads we did not build, eating foods we did not pick or raise. Allowing the illusion of my own independence to drop away unmasked a fundamental truth of being human [24]. This sentiment echoes throughout current disability studies: we are all interdependent, and independence has always already been an illusion.13 But disabled and old people have often needed to advocate for independence, because they have historically been the subject of laws, de facto rules, and public norms that deny them full citizenship, access, and opportunity to participate in public life. For example, early disability activism of the 1970s and 1980s produced a movement called the independent living movement. The name — independent living — indicates a reaction to the historical framing of disabled people as completely dependent, as objects to be moved around by nondisabled people. Despite the name, this movement was not about independence in the sense of isolation, total responsibility for one’s own needs, and total avoidance of inconvenience to nondisabled people — the values that underpin current narratives about AI. Instead, the movement was about the right to a public life, a life outside of institutions — an interdependent life. It demanded community living (the slogan “Our Homes, Not Nursing Homes!”), access to education and to public spaces (schools, grocery stores, shops, etc.), vocational services (to enable people to work), and assistive technologies (to help people move and remain outside of institutionalized and segregated care). The independent living movement demanded only that disabled people have the same autonomy, rights, and freedoms that nondisabled people enjoy. Interdependence (neither isolationist independence nor dependence) is a central value in the Crip Technoscience Manifesto. It is the second item on the list: “Crip technoscience is committed to interdependence as a political technology” [17, p. 12], and the Manifesto itself, which is “premised on interdependence,” was written to describe the “core values” of crip technoscience. According to Hamraie and Fritsch [16, 17], crip technoscience blends critical disability studies with approaches from the field of Science and Technology Studies (STS). It anticipates a world in which disabled people are seen as experts about their experiences and the materiality and context of their bodies. Crip technoscience anticipates accessible futures that work not toward cure or normalization but toward freedom. AI as a disability technology also presents particular problems of data privacy, especially given the long and ongoing history of the policing of disabled bodies. For example, the recently deployed Electronic Visit Verification (EVV) technologies enable personal care attendants (PCAs) to clock in and out of their home health visits. Under EVV, these personal care attendants, who provide physical support for disabled people to accomplish daily tasks (toileting, prepping food, getting in a wheelchair), are issued mobile phones to track their time on the job. But disability activists express worry about the scope of the monitoring enabled by this technology. As disability activist and filmmaker Dominick Evans tweeted: It is a pure violation of the civil liberties and personal freedom of disabled people across the United States. Our PCAs are now being equipped with something equivalent to an ankle bracelet by the US government! They are giving [sic] phones that have cameras in the front and the back so they can track exactly what all of us who need homecare are doing. They could record any conversation. These devices have to stay in our homes, at all times. We’ll never know when they are recording, and we don’t know who has access to them beyond the monitoring company, and government agencies. It’s a huge threat to the personal safety of disabled people who require home care [12] The use of EVV technology is a huge invasion of privacy — an invasion that would not be permitted for many segments of society.14 Can you imagine if any nondisabled, noncriminal citizen was required by the government to allow a device with the potential to monitor them in their homes at all times? This is precisely what some of the proposed AI technologies will do. And disabled people — especially disabled people who are multiply minoritized, whose identities are more carefully scrutinized by others — are rightfully concerned. Our society does not value the privacy of older people and disabled people,15 and many of the AIs in development sacrifice this privacy in favor of the peace of mind of presumably nondisabled caretakers — an assumption that disabled and old people are not agents in their own right.

Experiences of Real Disabled People

The crux of the problem is that AI designs, like other technologies meant to address disability issues, are often spearheaded by people not deeply embedded in the disability community —people who simply imagine what it might be like to be disabled or elderly rather than base their design on the experiences and needs of real disabled people. These designers, who are usually ignorant of the larger history of disability, often reinscribe ableism in their designs, further stigmatizing and marginalizing disabled people through monitoring or tracking or decision-making by proxy. The typical framing of new technologies as promoting independence is based on a lie: the lie that any of us is truly independent. These technologies are touted as empowering, but designing technologies without consulting their user base is the ultimate in disempowerment. One of the commitments of the Crip Technoscience Manifesto is to center “the work of disabled makers and knowers” [17, p. 7]. Tech designers’ reliance on their imagination of what it is to be disabled16 keeps the focus on individual functioning and limitation, rather than addressing the larger context — the poor infrastructure and social stigmas that work against the full participation of disabled people in society. This is not an either/or situation: we can both work to support individuals’ particular interests and needs and work to create a more accessible world for allpeople. But we cannot do either if we continue to design technologies that reinscribe the “cure” or normalization of individual disabled bodies and minds instead of making worlds more conducive to our existence.

Sign in to recut, update, or save this card.