|
@hypotext | |||||
|
Today, @theintercept published an exposé by Rodrigo Ochigame (a PhD candidate at MIT) on the invention of "ethical AI" by big tech.
theintercept.com/2019/12/20/mit…
Here's why I think it's important to read this piece and share it widely. 1/
|
||||||
|
||||||
|
Katherine Ye
@hypotext
|
20. pro |
|
Like many computer science researchers, I watched with confusion as the field of "fairness, accountability, transparency, and ethics in artificial intelligence" (FATE) was swiftly birthed in 2017. 2/
|
||
|
|
||
|
Katherine Ye
@hypotext
|
20. pro |
|
I don't work in the area, but I work adjacent to it. In the summer of 2017, I interned at Google Brain on visualizing machine learning. I saw FATE-focused initiatives, institutes, partnerships, and nonprofits spring up at major institutions after that summer. 3/
|
||
|
|
||
|
Katherine Ye
@hypotext
|
20. pro |
|
I saw many well-intentioned early grad students, as well as many senior faculty looking for a new cause, sucked into working in "ethics in AI."
Why was everyone all of a sudden working on this stuff? 4/
|
||
|
|
||
|
Katherine Ye
@hypotext
|
20. pro |
|
Given that algorithmic injustice has existed since, perhaps, the dawn of computing, I didn't think American computer science academics had suddenly started caring en masse—nor that algorithmic injustice was simply another "technical problem" that academics could solve. 5/
|
||
|
|
||
|
Katherine Ye
@hypotext
|
20. pro |
|
Regardless, in the last three years, work done under the aegis of promoting "fairness, accountability, transparency, and ethics" has become enormously influential in setting big tech and US policy agendas. 6/
|
||
|
|
||
|
Katherine Ye
@hypotext
|
20. pro |
|
Today, we find out that the language of "AI ethics" has been a smokescreen for corporate lobbying. Big tech groomed academia to develop technical self-regulations to avoid legal regulation. Why? Profit. Read Rodrigo's exposé for the receipts. 7/
theintercept.com/2019/12/20/mit…
|
||
|
|
||
|
Katherine Ye
@hypotext
|
20. pro |
|
Seriously, you won't find this amount of dishing anywhere else. Where few others would dare, Rodrigo has bravely come forth to illuminate the collusion between corporate, academic, and military pushers of "ethical AI." 8/
|
||
|
|
||
|
Katherine Ye
@hypotext
|
20. pro |
|
Do you care about how corporate profit agendas shape the research, policymaking, and national conversations around the software that governs your life and that is deployed against enemies of the state? Read this piece. 9/
|
||
|
|
||
|
Katherine Ye
@hypotext
|
20. pro |
|
This piece is a great counterpoint to the argument of technological determinism: "If we don't build it, someone else will." Perhaps interrogate who planted the thought that "it," whatever technology, was worth building in the first place? 10/
|
||
|
|
||
|
Katherine Ye
@hypotext
|
20. pro |
|
Finally: this is not new! In the 1970s and 80s, the US private insurance industry invented the idea of "actuarial fairness" to protect itself from charges of discrimination by activists and evade legal regulation, as Ochigame et al. have written earlier. Sound familiar? 11/
|
||
|
|
||
|
Katherine Ye
@hypotext
|
20. pro |
|
For more, check out their paper "Beyond Legitimation: Rethinking Fairness, Interpretability, and Accuracy in Machine Learning": 12/
dropbox.com/s/6ue5knrlvbxi…
|
||
|
|
||
|
Katherine Ye
@hypotext
|
21. pro |
|
Important note (via @rg9119):
twitter.com/fredbenenson/s…
|
||
|
|
||
|
Katherine Ye
@hypotext
|
21. pro |
|
I boosted this piece because it rings true with my + friends' experiences in CS academia & bc it starts an important conversation about influence + ideas. But I think it's important to note the tension b/t the author's intent with the piece and the impact it may actually have.
|
||
|
|
||
|
Katherine Ye
@hypotext
|
21. pro |
|
Seconding @fredbenenson's point that this is a complicated take, and to center voices from folks with other lived experiences in the conversation. twitter.com/fredbenenson/s…
|
||
|
|
||
|
Katherine Ye
@hypotext
|
21. pro |
|
A friend in the FAT* community adds that they think the piece flattens complex issues around incrementalism vs reform in the field, but that it's hard to have a public conversation about the field's struggles with power, because it requires disclosing even more insider knowledge.
|
||
|
|
||
|
Katherine Ye
@hypotext
|
22. pro |
|
Lots of complicated conversations happening now around this piece. I want to second @gleemie's call to center grassroots orgs that are led by the folks who are most affected by algorithmic injustice—as the writer does!
twitter.com/gleemie/status… pic.twitter.com/D80lMXQHtQ
|
||
|
|
||
|
Katherine Ye
@hypotext
|
22. pro |
|
And to focus on the structural forces at work in shaping the discourse around "ethics in AI." twitter.com/gleemie/status…
|
||
|
|
||