carolyn q zou
cqz [at] cs [dot] stanford [dot] edu
I'm a research scientist working on multiagent AI safety, and am currently based in San Francisco.
I am on leave from my PhD at Stanford, where I was advised by Michael Bernstein and Percy Liang, and was supported by the NSF GRFP and Stanford Graduate Fellowship. I spent some of this leave as the first hire at Simile.
My undergrad was in communication studies and philosophy, with a focus on semiotics. I was also a high school dropout.
Other interests include rhythm games, birding, and photography.
- Generative Agent Simulations of 1,000 People[link]
- Oak Story: Improving Learner Outcomes with LLM-Mediated Interactive Narrativespublished in uist '25. [link]
- LLM agents outperform Harvard negotiatorspresented at heal workshop (spotlight), chi '25.
- Validation Without Ground Truth? Methods for Trust in Generative Simulationspresented at heal (spotlight) & trew workshops, chi '24
- Sociotechnical Risks of Simulating Humans with Language Model Agentsundergraduate thesis at northwestern. [link]