Nature review maps four LLM roles transforming social science

Nature Human Behaviour has published a Review arguing that large language models now occupy four distinct roles in computational social science, and that the field will have to reckon with using the same technology as both tool and subject of study.

The Review, led by Fengli Xu, Chenyang Shao and colleagues including James Evans, surveys recent work across sociology, political science, psychology, anthropology, economics and related fields. LLMs, the authors write, now "serve four distinct research roles: subjects that simulate human behaviour, assistants that automate analysis, collaborators that generate hypotheses and evaluators that audit and improve research quality."

The framing is as striking as the taxonomy. "This transformation arrives as society itself becomes saturated with artificial intelligence agents," the authors write, "making LLMs both our research tools and authentic research subjects." They call it "a recursive challenge that will define the future of social science."

The paper does not claim the method is ready. "LLMs struggle with causal inference and raise urgent ethical concerns about privacy and misinformation," the authors write, "and their ability to reproduce social complexity remains uneven and unproven." The review stops short of naming a benchmark or reproducibility protocol for the four roles it maps.