Allison Duettmann Praises Double-Blind AI Evaluations
Allison Duettmann calls a piece on double-blind evaluations one case where cryptography aids AI safety.
TLDR
Allison Duettmann, President and CEO of Foresight Institute, posted praise for a piece on double-blind evaluations. She said it shows one way cryptography can ease tradeoffs between privacy and transparency in AI safety work. Duettmann congratulated @iamtrask, @wsisaac, @SolomonMg, @KLdivergence, @koenvanderveen, @bitsofsteve and @falcon_ronnie on the effort. A generated headline in the post references Google DeepMind piloting such evaluations for frontier AI.
Combined views
2.1K
1 Source, first seen 32d ago