Debating AI Skeptics: The Cognitive Shape of LLM Agents

JacquesThibs · x · 2026-08-19

Jacques Thibs responds to AI skeptics who doubt fast progress, arguing that the cognitive shape of LLM agents differs from human researchers. He suggests AI could offer 10,000x improvements in plotting and 1,000x in specific problem-solving, while only 0.5-10x in novel approaches. This distinct capability profile leads to a different pace of progress than simply scaling human researchers.

Original post →

More from AGI Musings

AGI Musings channel →