Lexin Zhou

Hi! I am a 1st-year CS PhD candidate at Princeton University, advised Prof. Peter Henderson at the POLARIS Lab. Before joining Princeton, I was a StarBridge Scholar at Microsoft Research for a year, working with Dr. Xing Xie. I did my master’s in CS at the University of Cambridge, funded by Open Philanthropy and supervised by Prof. Andreas Vlachos, and my BSc in Data Science at the Universitat Politècnica de València, where I got into research by working with Prof. Jose Hernandez-Orallo.

Over the past few years, my research interests have centered on evaluating AI capabilities, societal impact, and safety, with a special emphasis on general-purpose AI like LLMs, agents, and LLM-judges. I am a computer scientist by training but also regularly draw inspirations from interdisciplinary fields beyond CS, such as psychometrics and cognitive science.

At present, I mostly spend my days designing systematic evaluation frameworks that are not only robust, valid, efficient, and hard to game, but also enable us to causally explain and predict AI’s capabilities and limitations across a wide range of scenarios—rather than being constrained to arbitrary sets of narrow, task-specific benchmarks that lack explanatory and predictive power and are usually brittle, construct-invalid, inefficient and easy to cheat on. Recently, I have also become excited about a new training paradigm that is guided by principled, systematic evaluation frameworks, for building more reliable and safe AI systems. Whenever my availability allows, I also spend a good share of my time assessing and anticipating AI’s societal impacts (e.g. user-centric reliability, human-AI teaming) and ponder ways to make AI more socially beneficial.

I’ve spent time in research/consultancy roles at Microsoft Research, OpenAI, Meta AI, European Commission JRC, Krueger AI Safety Lab, and VRAIN. My work has been featured in Nature, Financial Times, Microsoft Research, MIT Tech Review, Forbes, IEEE Spectrum, El País, New Scientists, QbitAI, IBM, among others.

If you are drawn to everything relevant to AI Evaluation and wanna stay informed, please subscribe our monthly AI Evaluation Digest newsletter! If you wanna talk about something I do, feel free to reach out via email or on Twitter.