Junsol Kim
University of Chicago · Google

How artificial intelligence reshapes what we know, believe, and decide together.

I study the sociology of knowledge and artificial intelligence, focusing on how machines mediate the flow of information across ideological and cultural divides, how groups of humans and AI agents reason together, and how AI can expand the methodological toolkit of social science.

Ph.D. candidate in Sociology, University of Chicago · advised by James Evans
Ph.D. Student Researcher, Google, Paradigms of Intelligence
junsol@uchicago.edu · CV · Google Scholar · GitHub

Portrait of Junsol Kim

Featured

All projects
First page of Differential impact from individual versus collective misinformation tagging on the diversity of Twitter (X) information engagement and mobility Nature Communications 2025 First page of Linear representations of political ideology emerge in large language models ICLR (Oral) 2025 First page of Hidden persuaders: LLMs' political leaning and their influence on voters EMNLP 2024 First page of Biased AI improves human decision-making but reduces trust Working Paper Under review First page of AI-augmented surveys: Leveraging large language models and surveys for opinion prediction Working Paper Under review
01

Political and Cultural Divides in Knowledge Systems

How scientific knowledge and public information become distorted as they travel across ideological communities, from fact-checking on social platforms to the citation of science in partisan policy.

Read →
02

AI as Mediators and Agents of Knowledge

How large language models persuade, disagree, and reason: in conversations with humans, in networks with other agents, and through the political perspectives they silently encode.

Read →
03

AI for Social Science

Using generative AI to recover public opinion on issues never surveyed, to simulate human respondents across time and demography, and to surface systematic biases in whose views are predictable from text.

Read →

Publications

Full list

News

  • 2026.08
    Presenting Scientific misinterpretation and information laundering in policymaking at ASA 2026, New York City.
  • 2026.07
    Presenting Reasoning models generate societies of thought at IC2S2 2026, Burlington, VT.
  • 2026.06
    Presenting Scientific misinterpretation and information laundering in policymaking at ICSSI 2026, Boulder, CO.
  • 2026.05
    Talk at AI for Social Science Research Methods, Yale University: Reasoning models generate societies of thought.
  • 2025.11
    Selected as a Rising Star in Data Science (Stanford · UChicago · UCSD).
  • 2025.07
    Started as Ph.D. Student Researcher at Google, Paradigms of Intelligence Team.
  • 2025.04
    Oral presentation (top 1.8%) at ICLR 2025: Linear representations of political ideology emerge in large language models.
  • 2025.01
    Differential impact from individual vs. collective misinformation tagging published in Nature Communications.