AI science loops have too much textbook, not enough Newton-style curiosity
generativist · x · 2026-10-25
A pointed critique of current RL-driven automated science loops: they're overloaded with textbook knowledge and lack the adventurous, boundary-pushing questioning that made Newton go off to interrogate the occult.
More from AGI Musings
- Epoch AI: Memory Chips Shipped Through 2027 Could Run 33–171 Million Frontier AI Agents — PeterDiamandis · 2026-10-25
- 'Cope' meme thread mocks every common argument against AGI risk — tensorqt · 2026-10-25
- Leslie Lamport's "Claude, make no mistakes" hailed as predicting LLMs 30 years early — akbirthko · 2026-10-25
- A 3-year AI pause could buy 1000 years of safety work — or just 41, critics model — gleech · 2026-10-25
- Cambridge lecturer's decade-long poll: demand for 'Uber for personal security' near zero, now shifting — SamuelAlbanie · 2026-10-25
- With Claude writing code fast, where should engineers go next? — alejomos · 2026-10-25