Researcher: biological neurons are alive cells, LLMorphism is a fundamental AI mistake
ValerioCapraro · x · 2026-09-15
Behavioral scientist Valerio Capraro argues AI practitioners make a fundamental error by treating neurons as simple input-output units — the assumption ANNs are built on. A biological neuron is a living cell that regulates its internal state, synthesizes proteins, repairs itself and rewires — a property called autopoiesis. LLMs, he argues, are an extreme simplification of the brain: beware 'LLMorphism'.
More from AGI Musings
- Switching providers won't protect your data — build your own intelligence now — sarahookr · 2026-09-15
- Why would an OpenAI IPO make its AI-safety situation worse? Reddit asks — Stabile_Feldmaus · 2026-09-15
- Reddit thread calls Anthropic the most hypocritical AI lab on safety — Current_Balance6692 · 2026-09-15
- AI safety feud erupts: extinction claims of 10-70% vs fear of humans misusing AI — robleclerc · 2026-09-15
- Lila Sciences founder: building scientific superintelligence via self-play against reality — tw_killian · 2026-09-15
- AI safety debate: should worriers study intelligence itself, or existing systems? — aphercotropist · 2026-09-15