Sep 4, 2026 · 31 min
Jennifer Chayes maps the mathematics of massive networks
For Jennifer Chayes, math is a 'native tongue'
The episode connects abstract mathematics to the systems shaping technology, while asking what institutions and habits could make AI safer and more equitable.
- 1Graphons and phase transitions reveal collective behavior in enormous networks without tracing every individual interaction.
- 2Chayes’s unconventional path shaped a willingness to cross disciplines, tolerate failure, and pursue ambitious opportunities.
- 3AI could expand access to expertise and accelerate science, but only with verification, education, and institutions built for safety and equity.
Don't miss
Chayes explains how AI could accelerate science and widen access to expertise, while insisting that users learn to verify chatbot outputs and demand references.
The brief
Jennifer Chayes treats mathematics as a native language for making sense of the world, using phase transitions and graphons to explain how massive networks behave.
Graphons let researchers study the collective behavior of enormous networks without examining every connection, with applications ranging from social platforms to video interpretation.
Chayes’s path ran from dropping out of high school and living on New York City streets to an interdisciplinary research career spanning mathematics, science, and technology.
Her account of academia includes a retrospective recognition of gender bias and institutional barriers, even after she initially believed discrimination had largely passed her by.
Chayes sees generative AI as a potential engine for scientific discovery and broader access to expertise, but says people must question outputs, request references, and verify claims.
Her closing advice—seize opportunities despite difficult timing—lands through the story of accepting a major opportunity shortly after her mother’s death.
Featuring
Jennifer Tour Chayes
Microsoft Corporation
FacebookListen to the full episode and explore every guest, topic, and moment on PodLume.