Shane Legg and leading consciousness researchers release tool scoring LLMs 1%-80% on consciousness credence
mhutter42 · x · 2026-09-29
Shane Legg (DeepMind co-founder), Anil Seth, Chris Frith and collaborators have released a book-length pragmatic Bayesian treatment of consciousness plus an interactive tool for estimating the credence that a system is conscious. Current LLMs score between 1% and 80%. The team invites criticism, extensions and alternative assessments.
More from AGI Musings
- Does Exponential Compute Mean Exponential Intelligence? Tom Davidson and Ramez Clash Over Intelligence Explosion — connoraxiotes · 2026-09-29
- AI could force 11 million US workers into new careers by 2035 — cnn · 2026-09-29
- Post-training has a ceiling too: AI will match, not surpass, the best humans — Liu_eroteme · 2026-09-29
- Should the Ousterhout Dichotomy apply to AI models? Fixed tasks favor customized models — josh_wills · 2026-09-29
- Erik Hoel: physical systems have astronomically many description levels, none fundamentally different — erikphoel · 2026-09-29
- a16z on what would make an AI assistant worth paying for — a16z · 2026-09-29