LLMs aren't creating a new intellectual program — they're reviving symbolic AI's old questions
Amichayg · reddit · 2026-09-07
The author argues current AI discourse is overly LLM-centric, and many seemingly radical advances are older ideas becoming newly practical. Automated theorem proving via LLMs traces a lineage back to Principia Mathematica, the Logic Theorist, and the broader symbolic-AI tradition — foundational questions were posed early, then displaced by more scalable approaches.
Revisiting those traditions matters: the general concept of an automated proof assistant raises questions not exhausted by Lean, Agda, or whichever tool an LLM masters. Even if LLMs become excellent at operating these systems, that won't automatically surface the most distinctive interpretive questions — and the best frameworks may already exist in older traditions of logic, automated reasoning, and philosophy of mathematics. The current moment is less a new intellectual program, more old programs becoming viable.
More from AGI Musings
- a16z: World-changing tech makes a chaotic entrance — AI agents are following the same path — a16z · 2026-10-27
- NEJM AI on medical education: mastery requires performing without the machine — BraydonDymm · 2026-10-27
- How do you put odds on an AI apocalypse? Inside the p(doom) subculture — NathanpmYoung · 2026-10-27
- Dev warns abliterated models could wreck software supply chains with 1000+ transitive dependencies — nitarshan · 2026-10-27
- Delete your priors: why every SOTA model drop should reset how you think the world works — zack_overflow · 2026-10-27
- Replit predicts every Fortune 500 will be a 'Self-Driving Company' by end of 2030 — amasad · 2026-10-27