PeopleofSuperintelligence

Researcher

Chen Wu

Also known as 吴尘, Chen Henry Wu

Chen Wu is a Carnegie Mellon researcher whose work examines how language models learn from exploration, create diverse solutions, and verify reasoning. Born and raised in Nanjing, he studied linguistics at Tsinghua, worked with Yale's XLANG Lab, earned a computer-science master's at CMU, and began a PhD there in 2023. His research spans UnifiedSKG, language-model-generated robot rewards, multimodal-agent security, and creativity; his 2025 creativity paper won ICML's Outstanding Paper Award. In 2026, he coauthored Self-Trained Verification, and as of October 2026 he is a fourth-year CMU PhD student advised by Aditi Raghunathan, focused on model self-improvement, continual learning, and reasoning.

Timeline

  1. 2016–2020

    Linguistics at Tsinghua

    Born and raised in Nanjing, China, he studies linguistics at Tsinghua University from August 2016 to June 2020 and earns a B.A. in linguistics.

  2. June 2021–January 2022

    XLANG Lab research intern

    From June 2021 to January 2022, he works as a research intern with XLANG Lab and Yale University, with Professors Tao Yu and Dragomir Radev.

  3. January 2022–August 2023

    Graduate study at Carnegie Mellon

    He earns an M.S. in computer science at Carnegie Mellon from January 2022 to August 2023, then starts a PhD in computer science there in August 2023.

  4. 2022

    Unifying 21 knowledge tasks

    In 2022, he coauthors UnifiedSKG, a framework that unifies 21 structured knowledge-grounding tasks in a text-to-text format.

  5. 2024

    Reward code for robots

    In 2024, he coauthors Text2Reward, a framework that uses large language models to generate executable dense reward functions from natural-language goals for robotic manipulation and locomotion. The paper appears at ICLR 2024.

  6. 2024–2025

    Testing multimodal agents

    In 2024, he coauthors Dissecting Adversarial Robustness of Multimodal Language Model Agents, which introduces Agent Robustness Evaluation and 200 targeted tasks for web agents; the paper appears at NeurIPS 2024's Open-World Agents workshop and ICLR 2025. A single image covering less than 5% of webpage pixels can steer an agent toward a targeted goal with success rates up to 67%.

  7. 2025

    An ICML Outstanding Paper

    In 2025, he coauthors Roll the Dice & Look Before You Leap with Vaishnavh Nagarajan, Charles Ding, and Aditi Raghunathan, and the paper wins ICML's Outstanding Paper Award. Its minimal algorithmic tasks show that multi-token methods generate more diverse and original language-model output than next-token learning.

  8. As of October 2026

    PhD research at Carnegie Mellon

    In May 2026, he and Aditi Raghunathan introduce Self-Trained Verification, which trains a verifier to identify errors in generated reasoning by using reference solutions as supervision. As of October 2026, he is a fourth-year PhD student in Carnegie Mellon’s Robotics Institute, advised by Aditi Raghunathan. His research focuses on how language models explore and learn from exploration, and includes model self-improvement, continual learning, and reasoning.

Career

  1. 2021–2022Yale UniversityResearch Intern

Education

  1. 2023Carnegie Mellon UniversityM.S.
  2. 2020Tsinghua UniversityB.A. in linguistics

From the biography.

  1. 84% of people shaping AI with recorded undergraduate and doctoral schools changed institutions for their doctorateIncludes Yale University. 84% of people shaping AI with recorded undergraduate and doctoral schools changed institutions for their doctorate. Across 2,116 people with both an undergraduate degree and a doctorate recorded, we trace which undergraduate schools feed which doctoral programs, how often students stay at the same school, and how paths differ by birth-country region.
  2. People shaping AI work at a median of 3 organizations over a career: 3 for university starters and institute and lab starters, versus 2 for company startersIncludes Yale University. Across 6,853 people shaping AI with a dated work role, we estimate a median of 3 parent organizations over a career (mean 3.6), comparing role groups, starting sectors and starting decades from the 1950s to 2020–2026.
  3. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Yale University. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  4. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Yale University. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Chen Wu yet. Send us a free-licensed photo.

  1. Chen Wu chenwu.io. Checked 22 October 2026.
  2. UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language Models - ACL Anthology aclanthology.org. Checked 22 October 2026.
  3. Text2Reward: Reward Shaping with Language Models for Reinforcement Learning proceedings.iclr.cc. Checked 22 October 2026.
  4. Research - Chen Wu chenwu.io. Checked 22 October 2026.
  5. Dissecting Adversarial Robustness of Multimodal LM Agents proceedings.iclr.cc. Checked 22 October 2026.
  6. Roll the dice & look before you leap: Going beyond the creative limits of next-token prediction proceedings.mlr.press. Checked 22 October 2026.
  7. Self-Trained Verification for Training- and Test-Time Self-Improvement ar-forum.github.io. Checked 22 October 2026.
  8. Chen Wu - Robotics Institute Carnegie Mellon University ri.cmu.edu. Checked 22 October 2026.
  9. RI PhD Thesis Proposal - Chen Wu - Robotics Institute Carnegie Mellon University ri.cmu.edu. Checked 22 October 2026.