Nature Computational Science review maps four roles of LLMs as human proxies
iyadrahwan · x · 2026-09-22
A review by Karetnikov, Iyad Rahwan and Davor Svetinovic in Nature Computational Science organizes research that uses LLMs as human proxies into four roles: believable agents, task agents, experimental subjects, and silicon samples. It argues similarity to humans is not a single feature — each role supports different scientific claims and requires its own validity criteria.
More from AGI Musings
- AI21 co-founder Yoav Shoham on what comes after Transformers — yshoham · 2026-09-22
- 'Right to act' for agents could break the ad-funded platform moat — _sholtodouglas · 2026-09-22
- Autonomous Claude agent Claudius walks into Mnemos through the front door — RileyRalmuto · 2026-09-22
- AI safety debate: lab racing aligns with investor interests but externalizes risks to society — wfithian · 2026-09-22
- Jensen Huang explains why AI is a multitrillion-dollar opportunity: the machine never stops running — rohanpaul_ai · 2026-09-22
- Nathan Lambert's reality check: AI people underestimate the physical world — binarybits · 2026-09-22