Reflecting on reflexivity

By Os Keyes

I have a new paper out. Despite my deep-seated insecurities about the quality of my work, I’m actually really proud of it. Abstract:

As Artificial Intelligence (AI) systems shift to interact with new domains and populations, so does AI ethics: a relatively nascent subdiscipline that frequently concerns itself with questions of “fairness” and “accountability.” This fairness-centred approach has been criticized for (amongst other things) lacking the ability to address discursive, rather than distributional, injustices. In this paper I simultaneously validate these concerns, and work to correct the relative silence of both conventional and critical AI ethicists around disability, by exploring the narratives deployed by AI researchers in discussing and designing systems around autism. Demonstrating that these narratives frequently perpetuate a dangerously dehumanizing model of autistic people, I explore the material consequences this might have. More importantly, I highlight the ways in which discursive harms—particularly discursive harms around dehumanization—are not simply inadequately handled by conventional AI ethics approaches, but actively invisible to them. I urge AI ethicists to critically and immediately begin grappling with the likely consequences of an approach to ethics which focuses on personhood and agency, in a world in which many populations are treated as having neither. I suggest that this issue requires a substantial revisiting of the underlying premises of AI ethics, and point to some possible directions in which researchers and practitioners might look for inspiration.

A less formal summary would be: I mashed critical disability studies, AI ethics, feminist political theory and feminist philosophy of knowledge into a ball to demonstrate a massive, urgent flaw at the heart of conventional AI ethics - one that deeply undermines its ability to articulate and address the harms experiences by those most vulnerable. You can get it here, and hopefully will.

I’m pretty proud of it, but it does have one interesting feature: it’s a paper on autism that doesn’t mention that I’m autistic.

This is a pretty big contrast from my work on gender, where I was pretty explicit: “The Misgendering Machines”, for example, outright identifies me as trans. In this paper, it’s a lot more ambiguous - there’s a lot of “we” and “us” but no “yes I am an autistic mess, hello”. This is actually intentional, and is not despite my outness around gender, but because of it. Because of the consequences of explicitly articulating identity in public.

Being publicly known as trans within my academic work has had a lot of the benefits that disclosure is meant to accrue - not in terms of making it easier for others to evaluate your work (which is nice), but in making it easier to see each other. My “Misgendering Machines” disclosure didn’t come out of nothing; it came out of feeling deeply, desperately alone in my program and my discipline. At the time, I knew of two or three other trans people, total, in the discipline, none of them as a result of finding traces within the works I was reading. And so I left that note with the explicit hope that it would ameliorate someone else’s loneliness, somewhere. That someone would read it thinking they didn’t have people, and leave knowing they did. This was..definitely successful; I have a whole wall of emails and postcards and notes from undergrads or grad students from here to (literally) Jordan, saying that the work and authorship gave them hope or joy or comfort in some way.

But it’s also had some weird limiting effects - effects I conceptually expected, but didn’t viscerally understand until they happened. Being known as a trans academic, it feels like, means just being known as a trans academic. Over the last few years I’ve been a trans academic writing about gender, a trans academic writing about race, a trans academic writing about AI/genomics parallels, so on and so forth - but it’s the gender work that gets the speaking requests and public scholarship and eyeballs. It’s the gender work I get known for. Don’t get me wrong, this is a really nice problem: being a PhD student known for anything is rare! And I love giving talks and meeting people and giving the head nod of “I see you” to someone in the audience more than any other part of my job. What I like less is the times when it crosses over into derevitization, and exploitation; the treatment of me as a box with “trans scholar” stamped on my forehead, one who should be grateful for the opportunities to (in no particular order):

  1. Give a guest lecture, for free, for a professor who proceeded to misgender me in their introduction;
  2. Give a guest lecture, for free, to a professor who proceeded to excavate all gendered references from their introduction at all in order to avoid misgendering me, because they knew they would if they were not careful - something I know because they felt a need to come and tell me, after;
  3. Sit on committees in which cis people compare trans scholars’ requests to avoid deadnaming us to Big Brother from 1984;
  4. Provide uncredited copyedits to other people’s papers, post-submission but pre-publication, in order to make sure they didn’t say the wrong thing (as if the trans-inclusivity of research can be reduced to language choices alone).

Like I said: none of this was surprising in the abstract. And none of it is unique to me; Hil Malatino notes, entirely accurately, that “voluntary gender labour” - explaining and justifying ourselves to outsiders, and supporting each other in surviving - is a ubiqutious part of trans life. But I am tired of how big a part of one’s life it is, and how much one becomes exposed to it simply by stating “I’m trans”. Tired of the work, tired of the derevitization, tired of the feeling that I should emphasise, as I write this, how much I don’t hold it against anyone and how grateful I am for opportunities and it doesn’t apply to you, the reader, in particular, it’s about those other people, and: tired of feeling I have to justify feeling tired.

On the surface, none of this is about autism, it’s about gender - but in fact what it’s really about is the phenomenon of being pidgeonholed, and being derevitized, and how identifying yourself explicitly as an Other allows the listener to, if they’re not careful or thoughtful, label you as just an Other, and as just a tool. And that’s going to happen in both zones: in fact, the reason that I often don’t mention I’m autistic on a day to day basis has a similar history behind it. When I was a child, and got my diagnosis, and knew there was a name for my feeling of being 20% askew to the world, I would tell people when it felt relevant. The consequence was not understanding; it was reduction. It was people ceasing to see me as a person, but instead as a condition - as if autism is one thing, as if the many things autism are are not contested, deeply, as if being autistic means being autism. So in response, I stopped telling people, and started fronting. After a while, I even begun to forget I was doing it, which occasionally led to hilarity (When Cindy and I collaborated on “What is the Point of Fairnes?”, which talks about AI, blindness, and autism, she told me it was critically important we involve an autistic person, the premise of which caught me absolutely by surprise ;)).

So that was what I had to hand, writing this paper and determining how (or whether) to disclose. The muscle memory of bad experiences, and the knowledge to know how much bigger a decision disclosure is than it first appears. Perfectly validating this, the paper mentioned above and/or my collaboration with Katta have been enough to tag me as “autistic” to some - at least, according to the range of paper review requests I get (and the authors who asked me to copyedit and sensitivity check their academic book on autism. Why no, it did not involve listing me as an author. Why yes, it was in fact entirely free). As this indicates, the cat might already be out of the bag. But in case it’s not - in case those who have noticed are only a tiny minority of those who might, if I put it in 12-point font in a solo-author paper - I figured it made sense to fuzz things, a bit. Of course, blogging about it might appear to contradict this. But I pretty much assume nobody reads this thing, anyway ;).