PeopleofSuperintelligence

Researcher

David Rein

David Rein is technical staff at METR, a research nonprofit that measures whether and when AI systems might threaten catastrophic harm to society. He created GPQA, a benchmark of 448 graduate-level questions in biology, physics and chemistry, while a researcher at NYU. Before NYU, he was an early research engineer at Cohere, where he trained embedding models for semantic search. He also co-authored HCAST, a 189-task benchmark, and a randomized trial that found experienced open-source developers took 19% longer when allowed to use early-2025 AI tools. In January 2026, he discussed METR's time-horizon measurements on the AXRP podcast.

Timeline

  1. 2023

    The GPQA benchmark

    He creates GPQA while a researcher at NYU; its 448 graduate-level questions in biology, physics and chemistry reach arXiv in November 2023. Before NYU, he works as an early research engineer at Cohere, where he trains embedding models for semantic search.

  2. March 2025

    The HCAST benchmark

    He co-authors HCAST, a benchmark of 189 tasks in machine learning engineering, cybersecurity, software engineering and general reasoning, which reaches arXiv in March 2025.

  3. March 2025

    A 50% time-horizon metric

    He co-authors the paper “Measuring AI Ability to Complete Long Software Tasks”, which reaches arXiv in March 2025 and proposes a 50%-task-completion time horizon as a new metric.

  4. July 2025

    A developer productivity trial

    He is one of four authors of a randomized trial, which reaches arXiv in July 2025 and finds that experienced open-source developers take 19% longer when allowed to use early-2025 AI tools.

  5. January 2026

    Explaining time horizons on AXRP

    He discusses METR's time-horizon measurements as the guest of an AXRP podcast episode published in January 2026.

  6. As of October 2026

    Technical staff at METR

    He is technical staff at METR, a research nonprofit that scientifically measures whether and when AI systems might threaten catastrophic harm to society.

Sources and credits

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

  1. David Rein - METR metr.org. Checked 14 October 2026.
  2. GPQA: A Graduate-Level Google-Proof Q&A Benchmark arxiv.org. Checked 14 October 2026.
  3. HCAST: Human-Calibrated Autonomy Software Tasks arxiv.org. Checked 14 October 2026.
  4. Measuring AI Ability to Complete Long Software Tasks arxiv.org. Checked 14 October 2026.
  5. Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity arxiv.org. Checked 14 October 2026.
  6. 47 - David Rein on METR Time Horizons | AXRP - the AI X-risk Research Podcast axrp.net. Checked 14 October 2026.