David Baek's Website

☕️

David Baek

PhD Student

About Me

David is a PhD student at MIT EECS, where he is advised by Prof. Max Tegmark in the Tegmark AI Safety Group. His research interests include LLM Safety Post-training and Mechanistic Interpretability. In his free time, he enjoys playing piano and hiking.

Interests

LLM Safety Post-training
Mechanistic Interpretability

Education

PhD Computer Science
Massachusetts Institute of Technology
MS Computer Science
Massachusetts Institute of Technology
BS Physics & Computer Science
Seoul National University

📚 My Research

I am broadly interested in developing methods to improve the safety and reliability of LLMs. Please reach out to collaborate 😃

Featured Publications

Scaling Laws For Scalable Oversight

Large Language Models

Scaling Laws For Scalable Oversight

For the first time, we propose a framework that quantifies the probability of successful oversight as a function of the capabilities of the overseer and the system being overseen. We also find scaling laws in four different oversight games that approximate how domain performance depends on general AI system capability.

Apr 25, 2025

Physics-Inspired Effective Theory of Model Generalization

Learning Theory

Physics-Inspired Effective Theory of Model Generalization

We present GenEFT, an effective theory framework for shedding light on the statics and dynamics of neural network generalization, and illustrate it with graph learning examples.

Mar 20, 2025

Recent Publications

David Baek, Joshua Engels, Subhash Kantamneni, Max Tegmark (2025). Scaling Laws For Scalable Oversight. In NeurIPS 2025 (Under Review).

PDF Cite Code Custom Link

David Baek, Ziming Liu, Max Tegmark (2025). Physics-Inspired Effective Theory of Model Generalization. Physical Review E.