🧵1/ How should we balance accuracy and fairness when one predictive model has to serve multiple demographic groups? In our new paper with @annieulichney and Mike Jordan, “The Statistical Fairness-Accuracy Frontier”, we turn the fairness-accuracy frontier into a practical, finite-sample tool. TL;DR: Implementing fairness–accuracy trade-offs from finite data distorts the frontier; we provide minimax-optimal estimators, optimal sampling rules, and confidence bands for the empirical frontier.

We study a three-layer data market linking privacy-sensitive users, platforms, and a data buyer. We show that platform competition benefits data buyers more than users, and optimal privacy interventions depend on platform cost structures. @afallah94 econometricsociety.org/publi…

Professor Dimitri Bertsekas will be missed by many and many more to come… RIP.🙏🙏🙏

Graduate students & postdocs: the deadline to register for the poster session for the Crossroads of AI & Society Workshop is this Friday, May 29. Travel grants of up to $500 are available. Priority is given to poster presenters. Learn more and register at bit.ly/3RsWAta

Crossroads of AI & Society Workshop, hosted by Rice University's Global Paris Center in Paris, July 15-16, 2026. Speaker lineup across CS, operations and economics. Registration is free. Deadline: June 12 Poster and travel grant applications due: May 29 aaforml.com/2026/

The deadline for "Highlights Beyond EC" nominations is next week! Please submit work that you think would be socially valuable to put in front of the Econ&CS community. (Link is in replies.)

Our paper "Statistical Impossibility and Possibility of Aligning LLMs with Human Preferences" was just accepted to the Annals of Statistics. It's known that reward models can't represent Condorcet cycles. We make this *quantitative*, and prove a possibility result for Nash-based alignment. 📄 arxiv.org/abs/2503.10990 w/ Kaizhao, @DrQiLong , Zhekun, and @JiancongXiao14 1/n

📣 Registration is now open for the Crossroads of AI & Society Workshop, hosted by @RiceUniversity’s Rice Global Paris Center in Paris, July 15–16, 2026! 🇫🇷✨ 💡 Exploring the societal impacts of AI through the lenses of incentives, privacy, and fairness. 🎤 Featuring an amazing lineup of speakers across computer science, operations, and economics! 🎓 Poster session for grad students & postdocs! ✈️ Limited travel grants available! 🗓️ Registration is free; register by June 12. Apply for poster/travel support by May 29. Learn more register: aaforml.com/2026/ #AI #ML #ResponsibleAI #EconTwitter @RiceCompSci @RiceEngineering

Next up, we have Alireza Fallah (@afallah94) from Rice University, speaking about "Preference Alignment from Choices and Response Times".

This week’s workshop, “Agency in Collaborative Learning,” at @SimonsInstitute will feature many talks at the intersection of economics and machine learning. I’ll also be speaking on Tuesday about this work on preference alignment using choice and response-time data. The talks will be streamed online, and the schedule with the streaming links is available here: simons.berkeley.edu/workshop…

Congratulations to Moshe @Vardi, Univ Prof at Rice, named a 2026 NAAI Academy Award laureate by the National Academy of Artificial Intelligence. He was recognized for his seminal contributions to logic-based AI & formal reasoning in intelligent systems. bit.ly/4bv5ahg

If you’re on the Berkeley campus, I’ll be presenting this work at the @SimonsInstitute workshop on “Learning from Heterogeneous Sources” at 2:15 PM PT (less than an hour from now!). You can find the workshop schedule, along with the livestream links to all the talks (including mine), here: simons.berkeley.edu/workshop…

I'm happy to be joining a great cohort and grateful for the support of @coop_ai!

Did you just miss punching your ticket to Rio or Salt Lake City? Wanna go to a conference where people will engage with you and your paper on foundations of responsible computing, and you won't get lost in the crowd? Submit to #FORC2026, in Boston in June! Deadline in 2 weeks.

Do you work at the intersection of AI, mechanism design, strategic behavior, and their impact on society? Consider submitting your work to the “AI for Mechanism Design and Strategic Decision Making” (AIMS) workshop at ICLR 2026 (@iclr_conf). The submission deadline is February 3rd, and we welcome both short (4-page) and long (9-page) paper submissions. We also have a fantastic lineup of speakers! More information about submissions can be found on the workshop webpage: alimama-tech.github.io/aims-…

ICLR_2026_Workshops_names_20251201

2025 was an eventful/disruptive year for computer science research, for two reasons: 1) a shock to federal funding, and 2) the arrival of AI models capable enough to assist mathematical research. 1) is unambiguously bad and 2) is probably mostly good. I'll write about AI first.

I started a new personal blog The AI Observer on my observations on AI: kamalikachaudhuri.substack.c… Teaser: the first few posts will talk about generalization, alignment, what statistical learning theory gets wrong and right, and where the (non-obvious) open problems are buried.