Deb Raji Notes AI Fairness Ideas Recycled in Safety Discourse
UC Berkeley researcher flags repeated discourse cycles between AI fairness and safety communities.
Deb Raji posted replies on X expressing frustration that frontier AI safety work recycles ideas on auditing and labor from AI fairness discussions of about five years ago. She tagged Seth Lazar, Jessica Dai, Alondra Nelson, and the FAccT conference, urging better learning across groups rather than dismissal of prior communities like FAccT and MD4SG. Raji linked a 2019 ACM paper on fairness and abstraction, the MD4SG initiative started in 2016, and a 2023 piece on alignment. Seth Lazar replied that bridging these communities was a goal in 2023 and that such questions recur across philosophy and technology.
Been losing my mind lately at all the ways that the AI "fairness" discourse is effectively being recycled in AI "safety" discourse. Of course there are various things that are exciting & new problems but a lot of this is a surprisingly familiar cycle. Why not learn from the…
Reactions from ranked influencers
7 postsAnyways, here's some of that discourse from 2019; + Communities like MD4SG, who piloted in 2016; https://md4sg.com/aboutus.html And folks like @jessicadai_ explicitly drawing bridges to the language of "alignment" in 2023; https://joinreboot.org/p/alignment
@sethlazar @jessicadai_ @alondra @FAccTConference Oh this is not about territorialism at all - I just find the lack of communication across communities puzzling & inefficient. It seems lessons learnt on one side of the fence rarely makes it to the other side, and that just…
@sethlazar @jessicadai_ @alondra @FAccTConference Fwiw this is not a dig Ryan or Seb in particular - I think this shift is beyond welcome for the current "AI safety" discourse. It's just frustrating to watch the same cycle of discourse cycle through all over again, almost as…
@sethlazar @jessicadai_ @alondra @FAccTConference I also know this was something you wanted to intervene on in particular so I'm not sure if this is just me belatedly coming to terms with what you were trying to do circa 2023 in bringing the communities together to actually…
@sethlazar @jessicadai_ @alondra @FAccTConference All I know is that this is kind of driving me nuts - I notice folks working on "frontier AI" recycling discourse cycles from about 5 years ago (on auditing, labor, etc) while being dismissive of the contribution of communities…
Haha yes this was very much the goal back in 2023 :) but I also think that these are basically the fundamental questions about technology and society, so they’ve been reinvented any number of times over. I’m not sure the origins, probably Heidegger or something, or maybe even…