Landmark Nature paper: GPT-3 shows emergent analogical reasoning matching humans
ArtificialOther · x · 2026-10-26
Sharing the 2023 Nature Human Behaviour paper "Emergent analogical reasoning in large language models" (Taylor Webb, Keith J. Holyoak, Hongjing Lu). The study directly compared humans and GPT-3 (text-davinci-003) on analogical tasks including a non-visual matrix reasoning test based on Raven's matrices rule structure. GPT-3 showed surprisingly strong abstract pattern induction, matching or surpassing humans in most settings, zero-shot; preliminary GPT-4 tests were even better. With 489 citations and a 2025 Matters Arising follow-up, it remains a landmark in the debate over whether human-like cognition emerges in generic LLMs.
More from Research
- MemTensor's MemT-1.0-Nano trends on Hugging Face, a memory-tuned Qwen3.6-35B-A3B finetune — MemTensor · 2026-10-26
- Mathematicians dissect OpenAI's fluid singularity paper: unprecedented smooth-force construction — Singularitarian · 2026-10-26
- EMNLP 2026 Findings: "Ask or Assume?" turns coding agents into proactive help-seekers — najoungkim · 2026-10-26
- Subliminal learning transfers hidden traits across models without a shared base model — burny_tech · 2026-10-26
- Implementing SynthID-Text From Scratch: Watermarking Doesn't Hurt Output Quality — ell-hol1 · 2026-10-26
- Two teams independently discovered the same interpretability method—a sign the idea is robust — sedielem · 2026-10-26