Nature review sorts LLMs into four social-science roles
TL;DR
- A new Nature Human Behaviour review sorts LLM use in social science into four roles: subjects, assistants, collaborators and evaluators.
- The authors warn LLMs "struggle with causal inference" and that their ability to reproduce social complexity "remains uneven and unproven."
- Lead authors Fengli Xu and James Evans span fields from sociology to economics; the paper was published October 26, 2026.
A new review in Nature Human Behaviour sorts how researchers now use large language models into four roles: "subjects that simulate human behaviour, assistants that automate analysis, collaborators that generate hypotheses and evaluators that audit and improve research quality."
Published October 26, 2026 by Fengli Xu, James Evans and colleagues, the paper surveys sociology, political science, psychology, anthropology and economics.
The authors foreground the gaps. "LLMs struggle with causal inference and raise urgent ethical concerns about privacy and misinformation, and their ability to reproduce social complexity remains uneven and unproven," they write.
Evans trailed the review on X as a look at "where the field is fooling itself." The paper also flags a recursive wrinkle: LLMs are entering social science as research tools at the same moment society itself is becoming "saturated with artificial intelligence agents."
Two of the researchers we track posted the paper's link.
Shared on Bluesky by 2 AI experts
-
Large language models for computational social science www.nature.com/articles/s41...
View on Bluesky → -
Smart & helpful new paper in Nature Human Behavior on generative AI & social sciences which explores gen AI models as both novel tools for discovery and analysis, as well as objects of study - the latter thus esp insight…
View on Bluesky →
Originally reported by nature.com
Read the original article →Original headline: Large language models for computational social science - Nature Human Behaviour