Lecture Series in AI: Professors Micah Goldblum and John Hewitt @ColumbiaSEAS
Lecture Series in AI: Professors Micah Goldblum and John Hewitt  @ColumbiaSEAS
Uploaded April 2026 | Updated November 2026, 1 hour ago
The behavior of large language models---like those underlying ChatGPT---is not programmed; it is the emergent result of centuries of computation and an internet's worth of text. Understanding what goes on inside these language models is both one of the most exciting scientific opportunities, and one of the most pressing engineering challenges of our time. In this work, I'll provide an overview of this problem, and then discuss work my colleagues and I have done to discover what goes on inside these systems: do they learn the structure of human languages? How can we discover what concepts or ideas they have arrived at, which humans might not have thought of yet? And how can we make precision fixes for their problems?
Lecture Series in AI: Professors Micah Goldblum and John HewittSustainable Cloud Computing and AIBREAKOUT SESSION 1A: Columbia CryptoEconomics Working Session 2024Session 6: AI-Assisted Approaches to Data Collection and InferenceCCAE 2025: Decentralized Grids, Bolun XuTough Tech: EpiBoneSports AI Innovation Symposium | Lightning Talks 2: Sponsored Research PresentationsCEEC Symposium: The Grid Frontier - Lightning TalksReunion Lecture 2024. AI and CausalitySPLASHDOWN: Columbia engineers on Artemis II re-entryGotham Foundry Symposium: Regenerative Materials Across Industries Pt 1Future of Computational Sciences & Climate Modeling: Learning from the AI Revolution
Columbia Engineering |

Lecture Series in AI: Professors Micah Goldblum and John Hewitt

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER