As we shared in our earlier post introducing the Arena Academic Partnerships Program, Arena's mission is to measure and advance the frontier of AI through open, rigorous, and community-driven evaluation. Independent academic research is central to that mission, and today we're glad to introduce the first cohort of researchers supported through the program.
Congratulations to the first seven Arena Academic Fellows for Spring 2026:
Emmanuel Candès, Stanford University
- Statistically Efficient LLM Evaluation and Ranking with Pairwise Human Preference Data
Haifeng Xu, University of Chicago
- Towards a Principled Evaluation of LLMs’ Predictive Intelligence
Lihua Lei, Stanford University
- Is Bradley–Terry Enough? Heterogeneous Preference Modeling for RLHF
Negin Golrezaei, Massachusetts Institute of Technology
- Causal inference framework for multi-turn AI evaluation.
Ramesh Raskar, Massachusetts Institute of Technology
- Toward Adaptive, Task-Conditional Evaluation for AI Agents
Steven Wu and Andrew Ilyas, Carnegie Mellon University
- Goodhart’s Law on the Leaderboard: Can Cheap Preference Optimization Game Arena?
Tal Linzen, New York University
- A Rigorous Evaluation of Meta-Cognitive Monitoring in LLMs
We look forward to seeing this work take shape and to sharing updates as it progresses.
The Academic Partnerships Program will continue on an ongoing basis, with a call for proposals happening twice a year, each Spring and Fall. Learn more about how to submit a proposal for the Fall 2026 cohort here.
