Arguing consciousness may need complexity token-generation doesn't, even a tiny superhuman model wouldn't settle it
JoshPurtell · x · 2026-10-04
In a debate with Rob Bensinger and others, Josh Purtell argues that consciousness may rely on substantial neural complexity that behavioral thought — i.e., generating tokens — simply doesn't use. If so, even a tiny model (say 0.5B params) that cognitively dominates humans would tell us little about machine consciousness. He adds that the complexity needed to produce human behavior isn't obviously an upper bound or superset of what consciousness requires.
Related event: Machine Consciousness Debate: Is an LLM a Book or a Star-Sized Brain?(26 posts)→
More from AGI Musings
- AI pundit mocks Europe: 'bred not to desire too much' — same arrogance as mocking SpaceX — teortaxesTex · 2026-11-19
- AI alignment debate: a dark forest scenario where superintelligences preemptively strike GPU-hoarding rivals — jessi_cata · 2026-11-19
- OpenAI and Notre Dame Law School to co-host AI-for-humanity dialogue on Dec 4 — wjscheirer · 2026-11-19
- NYT: Anthropic Shows Organizational Fractures Ahead of Historic IPO — kylekabasares · 2026-11-19
- The biggest advantage of a top lab isn't money — it's saving you years — madeofAjala · 2026-11-19
- Ex-Anthropic staffer casts superintelligence as Sauron's ring; Ladish pushes back — JeffLadish · 2026-11-19