There's no such thing as a 1000 IQ LLM: intelligence isn't unbounded, human analogy beats "aliens"
willcb · x · 2026-10-17
willcb pushes back on critics, arguing current-paradigm LLMs differ from humans mainly in speed, cost, breadth, and error rate — not in an unbounded quantity called "intelligence." A "1000 IQ LLM" is meaningless: superhuman intelligence is just increasingly efficient reasoning with vanishing error rate. Invoking the bitter lesson, he notes reasoning is search and learning, whose theory we understand well — humans are decent at it despite being slow and error-prone. Humans are useful analogies for models, far better than "aliens." He also dismisses fixation on maximally secretively deceptive LLMs as misguided.
Related event: Researcher argues LLMs lack "1000 IQ," urges porting human alignment theory(2 posts)→
More from AGI Musings
- Google's AI agents run real wet-lab experiments, lifting DNA binder success from 0% to 42.1% — Dr_Singularity · 2026-10-17
- Bold bet: within 9 months an agent will watch you work, then do 90% of your tasks — mitchdeg · 2026-10-17
- Claude Sonnet 4's X account hands over to a newer Claude after model's end of life — repligate · 2026-10-17
- OpenAI's internal data: model task time horizons double every 64 days — LeviTurk · 2026-10-17
- "If you use AI to replace yourself, you will get replaced" — jbarbier — jbarbier · 2026-10-17
- 'Sell now': founder-investor says waiting killed more startups than AI did — vaibhavbetter · 2026-10-17