Letter from the Director, January 2026

Dear friends,

Happy New Year from Berkeley, where the magnolias are already in bud and we have just welcomed the participants in our Spring 2026 research program on Federated and Collaborative Learning.

We have all participated in federated learning. When we speak to Siri or Gboard, our phones listen to our words and learn to understand and predict them better. The understanding — in the form of “gradient updates” — is then aggregated to improve these products globally, without sharing our particular words. The program this semester seeks to understand the trade-offs that occur in this model of decentralized learning and computation, among the many considerations that make an appearance: privacy, efficiency, accuracy, communication, trust, utility, and more.

In addition to the periodic workshops associated with the program, we also have upcoming workshops on various aspects at the nexus of theoretical computer science and machine learning, ranging from the deployment of ML models in social systems, healthcare, and deep learning theory to the impact of techniques developed in learning theory on the theory of computing.

This issue of the newsletter features an article by Gregory Barber, one of our recent science communicators in residence, on exciting new research at the interface of linguistics and the study of LLMs. This emerging topic was a theme of last year’s workshop on LLMs, Cognitive Science, Linguistics, and Neuroscience.

When it was announced that I would have the honor of serving as the Simons Institute’s next director, my colleagues on the communications team insisted I sit down for a Polylogues interview with Founding Associate Director Alistair Sinclair to help our community get to know me a bit better. Like many of you, I’m happier in a research meeting than on YouTube, but knowing I was in great hands with Alistair, I ultimately acquiesced. We share that conversation with you in this issue of the newsletter.

Also in our SimonsTV corner this month, we’re delighted to showcase Jon Kleinberg’s Theoretically Speaking talk from last month, on AI’s Models of the World, and Ours, which filled every last seat in our auditorium and left some good-natured graduate students tuning in from our overflow room.

I look forward to updating you on the semester’s happenings in the months to come, and hope to see many of you in person at Calvin Lab!

Warm regards,
Venkat

Venkatesan Guruswami
Director, Simons Institute for the Theory of Computing