Yoav Goldberg: the class of things we assume AI can now do is kind of insane
teortaxesTex · x · 2026-11-03
AI researcher Yoav Goldberg notes that the class of things people casually assume "of course an AI can do that" has become kind of insane — e.g. expecting AI to compile neural network weights into a policy that plays a well-known Atari game, just because the technique appeared in many papers and undergrads could sketch a small version in an exam. Retweeted by teortaxesTex.
More from AGI Musings
- Psychics push back as AI-generated readings flood apps and TikTok Live — Chris_Armstrong · 2026-11-03
- Bending Spoons CEO: AI now writes 94% of our code, up from 0% in two years — illscience · 2026-11-03
- Lightcone Commons donor-routing mechanism draws candid caveat from evaluator — NathanpmYoung · 2026-11-03
- New data-constrained scaling laws point to quadrillion-parameter transformers, 7000x Chinchilla — teortaxesTex · 2026-11-03
- "Reddit is now 80% bots": dev laments AI-driven collapse of information ecosystems — vboykis · 2026-11-03
- Zvi rebuts Scott Alexander: case against SURA strategy is 'deeply confused' — TheZvi · 2026-11-03