AI People Wiki

Builders & open source

Maarten Sap

Assistant Professor, Language Technologies Institute (courtesy appointment in HCII), Carnegie Mellon University

Maarten Sap is an assistant professor in Carnegie Mellon University's Language Technologies Institute, with a courtesy appointment in the Human-Computer Interaction Institute, and a part-time senior research scientist and technical AI safety lead at the Allen Institute for AI (Ai2). Sap's research measures and improves the social and interactional intelligence of AI systems, assesses safety risks and socio-cultural biases in human- or AI-generated language, and builds narrative language technologies. Sap is a co-author of SOTOPIA, an open-ended environment in which language agents role-play social scenarios and are evaluated on their social intelligence.12

At a glance

Current roles
Assistant Professor, Language Technologies Institute (courtesy appointment in HCII) +1 more
Sector
University
Education
Carnegie Mellon, UW
Work listed
1
Honors
4
Sources cited
7
Profile reviewed

Career

Sap completed a PhD in computer science and engineering at the University of Washington from 2015 to 2022, advised by Yejin Choi and Noah Smith, with a thesis titled "Positive AI with Social Commonsense Models". During and after the PhD, Sap was a research intern, postdoctoral researcher and Young Investigator at the Allen Institute for AI, and interned at Microsoft Research in 2019.3

Sap joined Carnegie Mellon's Language Technologies Institute as an assistant professor in 2022 and has been affiliated faculty in the Human-Computer Interaction Institute since 2024. At Ai2, Sap's visiting role has grown from research scientist to senior research scientist and AI safety lead. At CMU, Sap teaches courses including Ethics, Safety, and Social Impact in NLP and LLMs.31

Research

Sap's doctoral work built resources for social commonsense reasoning. ATOMIC (2018) is an atlas of if-then knowledge about everyday events, SocialIQA (2019) is a benchmark for reasoning about social interactions, and Social Bias Frames (2019) models the social and power implications of language.456

Sap's group at CMU studies the social intelligence of AI systems, including theory of mind and pragmatic understanding, the safety of AI agents and people's reliance on them, cultural adaptability, and narrative understanding. SOTOPIA (2023) evaluates language agents as they role-play social scenarios, and "Artificial Hivemind" (2025) studies the homogeneity of open-ended language model outputs.127

Work and contributions

Timeline

  1. Began a PhD at the University of Washington advised by Yejin Choi and Noah Smith.3
  2. Posted ATOMIC, an atlas of machine commonsense for if-then reasoning, as first author.4
  3. Posted SocialIQA, a benchmark for commonsense reasoning about social interactions, as first author.5
  4. Posted Social Bias Frames as first author.6
  5. Became a postdoctoral researcher and Young Investigator at the Allen Institute for AI.3
  6. Joined Carnegie Mellon University's Language Technologies Institute as an assistant professor.3
  7. Co-authored SOTOPIA, an environment for evaluating social intelligence in language agents.2
  8. Co-authored "Artificial Hivemind", later selected for a NeurIPS 2025 Best Paper Award.73
  9. Became visiting senior research scientist and AI safety lead at the Allen Institute for AI.3

Honors and recognition

  • Packard Fellowship for Science and Engineering · David and Lucile Packard Foundation, 202531
  • Okawa Research Award · Okawa Foundation, 202531
  • NeurIPS Best Paper Award (Datasets and Benchmarks Track) · NeurIPS, 202531
  • ACL Outstanding Paper Award · Association for Computational Linguistics, 20233

Where they work

Suggest a correction or removal

Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.

References

7 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026

  1. Maarten Sap — homepageMaarten Sap · Official page · accessed Oct 21, 2026
  2. SOTOPIA: Interactive Evaluation for Social Intelligence in Language Agents (arXiv 2310.11667)arXiv · Paper · published Oct 18, 2023 · accessed Oct 21, 2026
  3. Maarten Sap — CVMaarten Sap · Official page · accessed Oct 21, 2026
  4. ATOMIC: An Atlas of Machine Commonsense for If-Then Reasoning (arXiv 1811.00146)arXiv · Paper · published Oct 31, 2018 · accessed Oct 21, 2026
  5. SocialIQA: Commonsense Reasoning about Social Interactions (arXiv 1904.09728)arXiv · Paper · published Apr 22, 2019 · accessed Oct 21, 2026
  6. Social Bias Frames: Reasoning about Social and Power Implications of Language (arXiv 1911.03891)arXiv · Paper · published Nov 10, 2019 · accessed Oct 21, 2026
  7. Artificial Hivemind: The Open-Ended Homogeneity of Language Models (and Beyond) (arXiv 2510.22954)arXiv · Paper · published Oct 27, 2025 · accessed Oct 21, 2026