Anders Sandberg: LLMs Show Rudiments of Theory Evolution but Fail at Constraints
anderssandberg · x · 2026-11-22
In a discussion with Grady Booch, Anders Sandberg argues LLMs show rudiments of evolving theories but are poor at maintaining constraints — the very reason symbolic reasoning took a while — and still lack good "taste" for theories.
cgcardona counters that the missing piece may not be symbolic reasoning itself, but the infrastructure for evolving theories: persistent structured state, provenance, and mechanisms for reconciling contradictory evidence.
Related event: Sandberg and Booch Debate Whether LLMs Can Evolve Theory(2 posts)→
More from AGI Musings
- Uber burned its annual AI budget in 4 months with no visible improvement — yaroslavvb · 2026-11-23
- Post-training may exceed 75% of FLOPs, undercutting perplexity-based safety bars — stochasticchasm · 2026-11-23
- FDA greenlights first human trial of age-reversal gene therapy — Dr_Singularity · 2026-11-23
- Gary Marcus doubles down: calling out overhyped LLM capabilities is engineering rigor — GaryMarcus · 2026-11-23
- AI safety researcher slams doxxing in the community: don't burn the commons on turn one — sjgadler · 2026-11-23
- AI proofs are a different kind of brute force: fast enough to skip inventing new math — moultano · 2026-11-23