Kerem Oktar

Postdoctoral Researcher,
Meta FAIR
[Fundamental AI Research]

šŸ“ƒ My CV

šŸ  Home, šŸ“§ Email
šŸŽ“ Scholar, šŸ¦ Twitter,
šŸŸ¦ Bluesky

Page Directory -> Homepage
-> Collaborators
-> Grad Application Advice
-> Stats Tutorials

The Science of

  • Social Intelligence.
  • Disagreement.
  • Persuasion.
  • Belief Change.

Social Intelligence in Minds and Machines

A secret to our success as a species is social intelligence: our ability to understand, learn from, and collaborate with one another. A corresponding limitation of modern artificial intelligence is its difficulty doing the same.

My research studies the mechanisms that underlie such social intelligence. I do this by building computational models (e.g., of how we detect persuasion) and testing these models with behavioral experiments in humans and AI. This research has two central pillars.

Understanding Human Social Intelligence. My research sheds light on basic puzzles of human social cognition. For example, why are people so good at learning from one another in some cases (e.g., scientific inquiry), yet so quick to ignore one another on important issues (e.g., political disagreement)? Our Paths to Persistence Model answers this through three interacting mechanisms (Psychological Review, 2025).

Engineering Artificial Social Intelligence. Second, I use these psychological insights to engineer better AI systems. Even frontier AI systems struggle with basic forms of social intelligence, such as detecting manipulative sources of information. Our research shows how Bayesian models of human social learning can be used to systematically improve these capacities in AI systems. (NeurIPS, 2025).

Who am I?

I was born and raised in Istanbul (šŸ§æ); studied economics and cognitive science at Pomona College, CA (ā˜€ļø); completed my PhD in Psychology at Princeton, NJ (ā„ļø); and am currently researching social cognition in Large Language Models at Meta in Seattle, WA (ā˜”).

Contact.

Feel free to contact me at oktar[dot]research[at]gmail.com with regards to research / collaboration / mentorship /ā€¦ - I love talking about science. If you would like to send me anonymous feedback, click here.

Resources.

I want science to be more inclusive and rigorous. Here are some resources that can help with that: