Stochastic parrots
Long, but worth reading — Tom Simonite, "What Really Happened When Google Ousted Timnit Gebru", Wired 6/8/2021.
The crux of the story is this paper, which is now available on the ACM's website: Emily Bender, Timnit Gebru, Angelina McMillan-Major, and Shmargaret Shmitchell, "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?🦜" In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, pp. 610-623. 2021.
As a result of a (somewhat strange) review process, described at length in the Wired article, Timnit Gebru and Margaret Mitchell were fired (or declared to have resigned) from their leadership roles in Google's Ethical AI group.
Read the rest of this entry »