Grace Liu

Hi I’m Grace. I’m a second-year PhD student in the MLD department at CMU. I research how to make RL and LLM agents more efficient, reliable, and interpretable. I am also interested in studying the challenges and opportunities of human-AI collaboration with increasingly capable AI agents at both the individual (cognitive) and societal scale. I am advised by Aarti Singh and work closely with Rachit Dubey and Benjamin Eysenbach. I am very grateful to have been advised by Tom Griffiths and Gabe Vecchi during my undergrad and Master’s at Princeton. I am also fortunate to be supported by the NSF GRFP fellowship.

Publications

AI Assistance Reduces Persistence and Hurts Independent Performance
Grace Liu, Brian Christian, Tsvetomira Dumbalska, Michiel A. Bakker, Rachit Dubey
Paper | Website

CaRT: Teaching LLM Agents to Know When They Know Enough
Grace Liu, Yuxiao Qu, Jeff Schneider, Aarti Singh, Aviral Kumar
Paper | Website

Demystifying the Mechanisms Behind Emergent Exploration in Goal-conditioned RL
Mahsa Bastankhah, Grace Liu, Dilip Arumugam, Thomas L. Griffiths, Benjamin Eysenbach
ICLR 2026, NYRL Workshop 2025 (Oral)
Paper | Website

A Single Goal is All You Need: Skills and Exploration Emerge from Contrastive RL without Rewards, Demonstrations, or Subgoals
Grace Liu, Michael Tang, Benjamin Eysenbach
ICLR 2025, IMOL Workshop @ NeurIPS 2024 (Oral)
Paper | Video | Code

Binary Climate Visuals Heighten Perceived Impact of Climate Change
Grace Liu, Jake Snell, Tom Griffiths, and Rachit Dubey
Nature Human Behavior
Paper | Code
Media coverage: Guardian, Grist, Gizmodo
Invited Op-eds: New Scientist

[*] denotes equal contribution