What if LLMs could decompose trained models into explicit ordinary code?
tszzl · x · 2026-10-17
A thought-provoking take on interpretability: following the bitter lesson, we (or our LLMs) might eventually decompose trained LLMs into explicit logical components expressible in ordinary code — an alternative to hand-reverse-engineered circuits where models explain and rewrite models. Pure speculation for now, no empirical results.
More from AGI Musings
- Friendly reminder: only ~2% of the population pays for AI today — xprunie · 2026-10-17
- Jensen Huang says AI will make us busier; the internet pushes back — VraserX · 2026-10-17
- Tech lawyer: outside SF, especially in Europe, people actively hate AI — and it's worrying — Scobleizer · 2026-10-17
- AI is pulling quantum timelines forward, but hardware is at the vacuum-tube stage — robleclerc · 2026-10-17
- Reviewers push back on Kevin Roose's AI book: 'white collar jobs are disappearing' claim called empirically untrue — Afinetheorem · 2026-10-17
- AI gives you a whole team of researchers, and some swear by it — Pascallisch · 2026-10-17