I am currently working within BAIR and CHAI, collaborating with Anca Dragan, Rohin Shah, and Jakob Foerster on developing frameworks for better collaboration between AI agents and humans. In the past I’ve worked with Yun Song on inferring tree topologies from DNA mutation data with deep learning methods.
My long term objective is to help reduce the risks associated with the use of new technologies, in particular ones related to AI and the Value Alignment problem.
On the Utility of Learning about Humans for Human-AI Coordination. Micah Carroll, Rohin Shah, Mark Ho, Tom Griffiths, Sanjit Seshia, Pieter Abbeel, Anca Dragan. NeurIPS 2019.
Workshop Papers and Other Writing
On the Utility of Learning about Humans for Human-AI Coordination. Micah Carroll, Rohin Shah, Mark Ho, Tom Griffiths, Sanjit Seshia, Pieter Abbeel, Anca Dragan. Imitation, Intent, and Interaction Workshop, ICML 2019.
Overview of Current AI Alignment Approaches. Micah Carroll. Final Project for AI Safety and Control (CS294-149 at UC Berkeley), 2018.
Machine Learning and Convex Optimization Class Notes
Sutton & Barto - Reinforcement Learning: Some Notes and Exercises
Summer 2018 - CS 188 Discussion Section