PeopleofSuperintelligence

Researcher

Margaret Li

Margaret Li is an AI researcher whose work spans dialogue systems, language-model training and evaluation. She earned a BAS in computer science at the University of Pennsylvania in 2017 and worked on dialogue research at Facebook AI Research. During her PhD at the University of Washington, she coauthored work on cross-document pretraining and the reproducibility of language-model scaling laws. Her coauthored Artificial Hivemind paper won a NeurIPS 2025 Best Paper Award, and as of October 2026 she is a final-year PhD student focused on LLM scaling and efficiency.

Badges

Timeline

  1. 2013–2017

    A computer science degree

    She earns a Bachelor of Applied Sciences in computer science with a mathematics minor at the University of Pennsylvania in 2017.

  2. 2017–2020

    Research engineer at FAIR

    From August 2017 to September 2020, she works as a research engineer at Facebook AI Research in New York on open-domain dialogue and reinforcement-learning agents for text-based games.

  3. 2020

    Reducing dialogue repetition

    In 2020, she coauthors Don’t Say That!, which extends unlikelihood loss to reduce repetition and logical flaws in dialogue generation.

  4. 2020–2024

    UW PhD and FAIR research

    She begins a PhD in computer science and engineering at the University of Washington in 2020, advised by Luke Zettlemoyer. From September 2020 to June 2021, she is a research scholar at Grid (now Lightning) AI Research; from October 2022 to October 2024, she is a visiting researcher at FAIR at Meta. In 2024, she coauthors In-Context Pretraining, which trains language models on related documents to support reasoning across document boundaries.

  5. 2025

    A survey of scaling laws

    In 2025, she coauthors (Mis)Fitting, an ICLR survey that finds many scaling-law studies under-report details needed to reproduce their findings.

  6. 2025

    A NeurIPS best paper

    She coauthors Artificial Hivemind, which wins a NeurIPS 2025 Best Paper Award and introduces Infinity-Chat, a dataset of 26,000 open-ended queries for studying language-model diversity.

  7. May 2026–September 2026

    FAIR research internship

    From May 2026 to September 2026, she works as a research scientist intern at FAIR at Meta.

  8. As of October 2026

    Final-year PhD student

    She is a final-year PhD student at the University of Washington’s Behavioral Data Science Group at the Paul G. Allen School, focused on language-model scaling and efficiency.

Career

  1. 2020–2021Lightning AIResearch scholar
  2. 2017–2026FAIR (Meta)Research engineer, Visiting researcher

Education

  1. From 2020University of WashingtonPhD
  2. 2017University of PennsylvaniaBachelor of Applied Sciences

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Margaret Li. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  2. 23% of people shaping AI who left a university for a company later returned to a universityIncludes Margaret Li. We estimate that 468 of the 2,011 people shaping AI who left a university for a company (23%) later returned to a university after a median of 5 years, and 163 of 829 (20%) made the reverse round trip.
  3. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Meta. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  4. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes Meta. Our estimates of the organizations where 550 people worked directly before joining 9 frontier AI labs across 620 career transitions, comparing the main feeder organizations per lab.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

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

  1. Margaret Li margs.li. Checked 21 October 2026.
  2. Don’t Say That! Making Inconsistent Dialogue Unlikely with Unlikelihood Training aclanthology.org. Checked 21 October 2026.
  3. In-Context Pretraining: Language Modeling Beyond Document Boundaries proceedings.iclr.cc. Checked 21 October 2026.
  4. (Mis)Fitting Scaling Laws: A Survey of Scaling Law Fitting Techniques in Deep Learning proceedings.iclr.cc. Checked 21 October 2026.
  5. Announcing the NeurIPS 2025 Best Paper Awards blog.neurips.cc. Checked 21 October 2026.
  6. Behavioral Data Science · University of Washington behavioral-data.github.io. Checked 21 October 2026.