Thinking of AIs as "guys in computers" beats "next token predictors"
typewriters · x · 2026-08-31
Melanie Mitchell retweeted Boaz Barak's discussion on using metaphors to understand AI behavior. Barak advocates for a pragmatic approach, suggesting that thinking of AIs as "guys living in computers" is more useful than parroting "they are just next word predictors." While AIs are not humans, many intuitions from human behavior carry over. Mitchell added a caveat: when human intuitions are directly mapped to agent behavior, it obscures technical specifics. While this has its place, it is less suitable for technically grounded discourse.
More from AGI Musings
- 'Interp was mostly hopeless before — now it's automatable': Cammarata on interpretability — burny_tech · 2026-10-16
- Inference engine team says they're in a real RSI loop: faster engine helps build faster engine — cephaloform · 2026-10-16
- FDA pays small company to build a grading system for AI-written radiology reports — YvesMulkers · 2026-10-16
- 'If we developed AI in the 1800s, people would demand ASI to cure diabetes' — zetalyrae · 2026-10-16
- AI trained on the fraction of human intelligence that fit the page — r0ck3t23 · 2026-10-16
- Alignment researcher Turn_Trout blasts AI assistants scraping private Signal chats without consent — Turn_Trout · 2026-10-16