PeopleofSuperintelligence

Researcher

Adina Williams

Adina Williams co-writes datasets and benchmarks for evaluating AI models. She completed a BS at the University of Michigan in 2010 and a master's in 2013 before joining NYU's linguistics PhD program. Her MultiNLI corpus, a sentence-understanding dataset she co-wrote, was published in 2018, and she co-authored the Dynabench platform in 2021. The PRISM Alignment Dataset, which she co-authored, won a Datasets and Benchmarks best paper award at NeurIPS. Listed with FAIR at Meta, she co-authors Calibrating LLM Judges, a 2026 ACL paper.

Timeline

  1. 2010

    A BS at Michigan

    In 2010 she completes a BS at the University of Michigan, then an MA at Michigan State University.

  2. 2013

    Starting an NYU PhD

    Her master's finishes in 2013 as she joins NYU's linguistics PhD program.

  3. 2018

    The MultiNLI corpus

    With Nikita Nangia and Samuel R. Bowman, she co-writes MultiNLI, a corpus for sentence understanding published in 2018.

  4. 2021

    The Dynabench benchmarking platform

    In 2021 she co-authors Dynabench, an open-source platform for dynamic dataset creation and model benchmarking.

  5. June 2024

    Measuring geographic disparities

    Posted to arXiv in June 2024, Decomposed evaluations of geographic disparities in text-to-image models lists her as a co-author.

  6. 2024

    A NeurIPS best paper award

    The PRISM Alignment Dataset, which she co-authors, appears on arXiv in April 2024 and wins a Datasets and Benchmarks best paper award at NeurIPS.

  7. June 2025

    The IntPhys 2 benchmark

    She co-authors IntPhys 2, a benchmark for intuitive physics understanding, which arXiv posts in June 2025.

  8. Since 2025

    Calibrating LLM judges

    Listed with FAIR at Meta, she co-authors Calibrating LLM Judges, a paper in ACL 2026's Industry Track.

Sources and credits

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

  1. Adina Williams, BS, 2010 | U-M LSA Asian Languages and Cultures prod.lsa.umich.edu. Checked 14 October 2026.
  2. [1704.05426] A Broad-Coverage Challenge Corpus for Sentence Understanding through Inference arxiv.org. Checked 14 October 2026.
  3. A Broad-Coverage Challenge Corpus for Sentence Understanding through Inference - ACL Anthology aclanthology.org. Checked 14 October 2026.
  4. Dynabench: Rethinking Benchmarking in NLP - ACL Anthology aclanthology.org. Checked 14 October 2026.
  5. [2104.14337] Dynabench: Rethinking Benchmarking in NLP arxiv.org. Checked 14 October 2026.
  6. [2406.11988] Decomposed evaluations of geographic disparities in text-to-image models arxiv.org. Checked 14 October 2026.
  7. Decomposed evaluations of geographic disparities in text-to-image models arxiv.org. Checked 14 October 2026.
  8. NeurIPS 2024 Awards neurips.cc. Checked 14 October 2026.
  9. [2404.16019] The PRISM Alignment Dataset: What Participatory, Representative and Individualised Human Feedback Reveals About the Subjective and Multicultural Alignment of Large Language Models arxiv.org. Checked 14 October 2026.
  10. The PRISM Alignment Project: What Participatory, Representative and Individualised Human Feedback Reveals About the Subjective and Multicultural Alignment of Large Language Models arxiv.org. Checked 14 October 2026.
  11. IntPhys 2: Benchmarking Intuitive Physics Understanding In Complex Synthetic Environments arxiv.org. Checked 14 October 2026.
  12. [2506.09849] IntPhys 2: Benchmarking Intuitive Physics Understanding In Complex Synthetic Environments arxiv.org. Checked 14 October 2026.
  13. Calibrating LLM Judges: Linear Probes for Fast and Reliable Uncertainty Estimation arxiv.org. Checked 14 October 2026.
  14. [2512.22245] Calibrating LLM Judges: Linear Probes for Fast and Reliable Uncertainty Estimation arxiv.org. Checked 14 October 2026.
  15. Calibrating LLM Judges: Linear Probes for Fast and Reliable Uncertainty Estimation - ACL Anthology aclanthology.org. Checked 14 October 2026.