PeopleofSuperintelligence

Researcher

Katie Millican

Katie Millican is an AI researcher whose work spans large language models, multimodal systems, and data. She contributed to Gopher, Chinchilla, and Flamingo, then co-led data for Gemini 1.0 in 2023. Her 2024 work included Gemini 1.5 and Gemma, and in 2025 she coauthored an IJCAI demonstration on code-generating strategies through self-play.

Works
12
Citations
4,947
h-index
8

Publication numbers from OpenAlex, as of 15 October 2026.

Timeline

  1. December 2021

    Gopher and model scaling

    In December 2021, she coauthors the Gopher study of language models, which evaluates systems on 152 tasks and includes a 280-billion-parameter model.

  2. March 2022

    Chinchilla and compute scaling

    In March 2022, she coauthors the Chinchilla study, which finds that model size and training tokens should scale together under a fixed compute budget.

  3. April 2022

    The Flamingo visual model

    In April 2022, she is among the contributors to Flamingo, Google DeepMind's visual language model for few-shot image, video, and text tasks.

  4. December 2023

    Co-leading Gemini data

    In December 2023, the Gemini report lists her as co-lead for data on Gemini 1.0, Google's multimodal model family.

  5. March 2024

    Gemini 1.5 and long context

    In March 2024, she coauthors Gemini 1.5, a multimodal model family designed to reason over millions of tokens of context.

  6. March 2024

    Gemma open models

    In March 2024, she coauthors Gemma, a family of open models built from the research and technology used to create Gemini.

  7. 2025

    Code strategies through self-play

    In 2025, she coauthors an IJCAI demo that uses large language models and self-play to generate code-based strategies for multiplayer games.

Sources and credits

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

  1. Scaling Language Models: Methods, Analysis & Insights from Training Gopher arxiv.org. Checked 15 October 2026.
  2. Training Compute-Optimal Large Language Models arxiv.org. Checked 15 October 2026.
  3. Tackling multiple tasks with a single visual language model — Google DeepMind deepmind.google. Checked 15 October 2026.
  4. Gemini: A Family of Highly Capable Multimodal Models arxiv.org. Checked 15 October 2026.
  5. Gemini: A Family of Highly Capable Multimodal Models arxiv.org. Checked 15 October 2026.
  6. Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context arxiv.org. Checked 15 October 2026.
  7. Gemma: Open Models Based on Gemini Research and Technology arxiv.org. Checked 15 October 2026.
  8. Combining Code Generating Large Language Models and Self-Play to Iteratively Refine Strategies in Games ijcai.org. Checked 15 October 2026.