Jennifer Tour Chayes is a distinguished mathematician and computer scientist whose research has spanned statistical physics, graph theory, and the theoretical foundations of machine learning. As of 2025, she serves as the Dean of the School of Information at the University of California, Berkeley, where she also holds appointments in the departments of Electrical Engineering and Computer Sciences and Statistics. Her career has bridged academia and industry, including a long tenure at Microsoft Research, where she co-founded and led the Theory Group and later the Microsoft Research New England and New York City laboratories.
Chayes earned her undergraduate degree in mathematics and physics from Wesleyan University and her Ph.D. in mathematical physics from Princeton University. Her early research focused on phase transitions and random graphs, contributing to the rigorous understanding of phenomena in statistical mechanics. This work laid a foundation for later contributions to computer science, particularly in the analysis of algorithms and the behavior of complex networks.
Academic and Research Career
Before joining Microsoft, Chayes was a professor of mathematics at the University of California, Los Angeles, and earlier held positions at the Massachusetts Institute of Technology and the University of Washington. Her academic work in the 1980s and 1990s included influential papers on the geometry of random surfaces and the dynamics of spin glasses, often in collaboration with her husband, Christian Borgs. Their joint research on the "free boundary" problem for random cluster models helped clarify connections between percolation theory and phase transitions.
Leadership at Microsoft Research
Chayes joined Microsoft Research in 1997 as a senior researcher and quickly became a Distinguished Scientist. In 2008, she co-founded the Microsoft Research New England laboratory in Cambridge, Massachusetts, serving as its managing director. Two years later, she helped establish the Microsoft Research New York City lab, focusing on computational social science and data science. Under her leadership, these labs became known for interdisciplinary work combining computer science with economics, sociology, and biology. She also played a key role in Microsoft's early machine learning initiatives, including projects on large-scale data analysis and algorithmic game theory.
Contributions to Machine Learning and AI
Chayes's later research shifted toward the theoretical underpinnings of artificial intelligence and deep learning. She has published on the generalization properties of neural networks, the geometry of loss landscapes, and the design of efficient algorithms for training large models. Her work on the "information bottleneck" method, developed with Naftali Tishby and Bill Pol, provided a framework for understanding how neural networks compress input data while preserving relevant information. This line of research has influenced subsequent studies on interpretability and robustness in modern large language models.
Berkeley School of Information
In 2023, Chayes was appointed Dean of the Berkeley School of Information, succeeding AnnaLee Saxenian. In this role, she has emphasized the importance of integrating technical rigor with policy and ethical considerations in the development of generative AI. She has advocated for interdisciplinary curricula that train students to address societal challenges such as misinformation, algorithmic bias, and data privacy. Under her leadership, the school has expanded partnerships with industry and government, including collaborations with Google Cloud and Amazon Web Services on responsible AI research.
Honors and Recognition
Chayes is a Fellow of the American Association for the Advancement of Science, the Institute of Mathematical Statistics, and the Association for Computing Machinery. She has received the Lady Davis Fellowship and the Humboldt Research Award. In 2019, she was elected to the National Academy of Engineering for her contributions to the theory and practice of data science and machine learning. She has also served on advisory boards for OpenAI and Google DeepMind, reflecting her standing as a bridge between theoretical research and applied AI development.
Selected Publications and Influence
Among her most cited works are papers on the phase transition in the random graph model, the stability of the internet's graph structure, and the spectral properties of adjacency matrices in complex networks. Her 2015 paper "The Information Bottleneck Method for Deep Learning" has been cited thousands of times and is considered a foundational contribution to the study of representation learning. She has also written extensively on the challenges of scaling machine learning systems, including issues of energy efficiency and the need for new hardware architectures, topics that remain central to debates about sustainable AI.
Chayes continues to publish actively and mentor early-career researchers. Her career exemplifies the productive intersection of mathematics, computer science, and public policy, and her leadership at Berkeley is shaping the next generation of information professionals.
Infobox
- Born: 1956 (exact date not publicly confirmed)
- Nationality: American
- Known For: Information bottleneck method, phase transitions in random graphs, leadership at Microsoft Research
- Affiliation: University of California, Berkeley; Microsoft Research (former)
- Awards: Fellow of AAAS, IMS, and ACM; Member of National Academy of Engineering; Humboldt Research Award