PeopleofSuperintelligence

Researcher

Anson Ho

Anson Ho is a senior researcher at Epoch AI, where he studies AI progress and its societal effects. He grew up in Hong Kong and studied physics in the UK. His 2022 and 2024 studies examine training-data growth and algorithmic progress in language models. In 2025, he joined the writing group of the International AI Safety Report, coauthored GATE, and coauthored a statistical framework for comparing AI benchmarks. He joined the writing group for the second International AI Safety Report in 2026.

Now at
Epoch AI
Works
18
Citations
671
h-index
8

Publication numbers from OpenAlex, as of 18 October 2026.

Timeline

  1. 2022

    Tracking training data

    He grows up in Hong Kong and studies physics in the UK. His 2022 work with Pablo Villalobos estimates training-dataset trends across more than 200 machine-learning models.

  2. March 2024

    Measuring language model progress

    In March 2024, he coauthors a study estimating that improvements in pretrained language models occur at a pace equivalent to doubling computational power every five to 14 months.

  3. November 2024

    Building FrontierMath

    The November 2024 paper introduces FrontierMath with Ho among its coauthors; the benchmark uses original mathematics problems vetted by experts to evaluate advanced mathematical reasoning in AI.

  4. January 2025

    Writing the safety report

    For the first International AI Safety Report, he joins its writing group in January 2025; the report reviews research on general-purpose AI capabilities and risks.

  5. March 2025

    Modeling AI automation

    In March 2025, he coauthors GATE, an integrated model that simulates the economic effects of AI automation.

  6. November 2025

    Comparing AI benchmarks

    A November 2025 paper coauthored by Ho introduces a statistical framework that places model capabilities and benchmark difficulties on one scale.

  7. February 2026

    A second safety report

    Ho joins the writing group for the February 2026 International AI Safety Report, a review of current research on the capabilities and risks of general-purpose AI.

  8. As of October 2026

    Research at Epoch AI

    As of October 2026, he is a senior researcher at Epoch AI, where he studies future AI developments and their societal impacts.

Career

  1. NowEpoch AISenior researcher
  2. 2013–2017University of Waterloo

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. Compute Trends Across Three Eras of Machine Learning2022 · 2022 International Joint Conference on Neural Networks (IJCNN) · 358 citationsWith Tamay Besiroglu
  2. Toward Transparent AI: A Survey on Interpreting the Inner Structures of Deep Neural Networks2023 · IEEE Conference on Secure and Trustworthy Machine Learning (SaTML) · 121 citations
  3. Will we run out of data? Limits of LLM scaling based on human-generated data2022 · 79 citationsWith Tamay Besiroglu
  4. Machine Learning Model Sizes and the Parameter Gap2022 · 37 citationsWith Tamay Besiroglu
  5. Modeling and Predicting Mobile Phone Touchscreen Transcription Typing Using an Integrated Cognitive Architecture2017 · International Journal of Human-Computer Interaction · 17 citations

From OpenAlex, as of 18 October 2026.

  1. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes University of Waterloo. Our estimates of where the people shaping AI studied and where their careers led, from education records in cited biographies.
  2. The company share of AI researchers’ top papers rose from 3% to 32%Includes University of Waterloo. We estimate where leading AI researchers worked when their most-cited papers appeared. The company share rose from 3% in 1975–1979 to 32% in 2020–2024.
  3. Google has the highest combined h-index among AI organizationsIncludes University of Waterloo. Our estimates of combined and median h-index at AI organizations, using current affiliations and OpenAlex records.
  4. Stanford University trained or employed 27 of the 111 AI major-prize winners, more than any other universityIncludes University of Waterloo. Stanford University trained or employed 27 of the 111 AI major-prize winners, more than any other university. We compare doctorate and employment links for major-prize and best-paper winners.

Colleagues

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

Sources and credits

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

  1. Anson Ho ansonwhho.github.io. Checked 18 October 2026.
  2. Trends in training dataset sizes epoch.ai. Checked 18 October 2026.
  3. Algorithmic progress in language models epoch.ai. Checked 18 October 2026.
  4. FrontierMath: A Benchmark for Evaluating Advanced Mathematical Reasoning in AI arxiv.org. Checked 18 October 2026.
  5. International AI Safety Report 2025 internationalaisafetyreport.org. Checked 18 October 2026.
  6. GATE: An Integrated Assessment Model for AI Automation arxiv.org. Checked 18 October 2026.
  7. A Rosetta Stone for AI Benchmarks arxiv.org. Checked 18 October 2026.
  8. International AI Safety Report 2026 internationalaisafetyreport.org. Checked 18 October 2026.
  9. Anson Ho | Epoch AI epoch.ai. Checked 18 October 2026.