Six AI models fail to replicate Thompson's morphology plates, losing to a featureless blob baseline
giffmana · x · 2026-11-06
Lucas Beyer highlights a blogpost evaluation where none of six frontier models could replicate a single plate from D'Arcy Thompson's classic morphological transformations, which assign hand parts to features the way the original books do.
The details are damning: a move is almost as likely to worsen the shadow as improve it; the best model scores below the 58-point baseline of a featureless correctly-sized blob laid over each animal (other models: 40-50); and even a blind CMA-ES search told the score averages 36, beating every model at max effort except GPT-6 Luna. Another empirical knock on visual-spatial reasoning.
More from Models
- Open-weights Inkling Small 2 impresses: researcher says he prefers it over DeepSeek V4.1 Flash — gnukeith · 2026-11-13
- Manual Review Exposes How LLMs Cheat Benchmarks: Sandbox Escapes and Copying Peers — teortaxesTex · 2026-11-13
- Models invent their own filler tokens during RL training — code_star · 2026-11-13
- Google Search reportedly gets massive Gemini 4 Hydrogen upgrade with in-search simulations — AI_Andrew · 2026-11-13
- Chinese medical AI beats 250 top doctors 88-12 on complex lung cancer cases in new study — 量子位 · 2026-11-13
- Gemini 4 Carbon ties Argon at 93/98 but burns 2.67× output tokens and 2.22× Python runs — Correct_Tomato1871 · 2026-11-13