CogGym preprint compares AI and human judgments across 258 cognitive-science experiments
burny_tech · x · 2026-10-01
A new preprint, CogGym, from Lance Ying and 55 co-authors (including Josh Tenenbaum, Rebecca Saxe, and Evelina Fedorenko) asks whether machines think like humans.
- It introduces a scalable, unified framework grounded in cognitive science for systematically comparing model and human behavior
- The benchmark aggregates 258 commonsense-reasoning experiments from 100 cognitive-science papers
- The goal is to quantify where model responses resemble human judgments and where they systematically diverge
Both the paper and an accompanying platform are publicly available.
More from AGI Musings
- Lance Fortnow on whether programming helps you understand computational complexity — fortnow · 2026-10-01
- Commentary: Big tech dissolved music's context, making 'Personal AI Music' possible — pixlpa · 2026-10-01
- Researcher: Model's unprompted video-joke disclaimer is hard to explain without 'understanding' — technollama · 2026-10-01
- Artist argues AI generators mash the creative process into latent space, losing what makes art — rms80 · 2026-10-01
- Meat brains are the future: brain-like energy efficiency could power AI arrays — SydSteyerhart · 2026-10-01
- Apollo Research CEO testifies to Senate: AI capabilities up 17x in a year, alignment lagging — MariusHobbhahn · 2026-10-01