PeopleofSuperintelligence

Researcher

Shane Legg

Shane Legg is a DeepMind cofounder and Google DeepMind's Chief AGI Scientist. He leads Google DeepMind’s AGI Safety Council. In 2007, he and Marcus Hutter published work on definitions of intelligence and a proposed mathematical measure of it. He founded DeepMind with Demis Hassabis and Mustafa Suleyman in 2010. He coauthored the 2015 deep Q-network paper and conducted 2017 experiments in learning from human preferences. More recently, he helped develop the Levels of AGI framework and coauthored a public safety approach. Those documents propose ways to describe capabilities and prepare for risks. Their statements about when AGI might arrive are forecasts.

Works
96
Citations
34,772
h-index
26

Publication numbers from OpenAlex, as of 14 October 2026.

Photo of Shane Legg

Timeline

  1. 2007

    Collecting definitions

    Legg and Marcus Hutter published a collection of definitions of intelligence in 2007.

  2. 2007

    A mathematical proposal

    Their Universal Intelligence paper proposed a formal measure based on an agent's ability to achieve goals across environments.

  3. 2010

    Cofounding DeepMind

    Legg founded DeepMind with Demis Hassabis and Mustafa Suleyman in 2010.

  4. 2023–2025

    Levels of AGI

    Legg coauthored Levels of AGI, first posted in 2023 and later revised.

  5. Present

    An internal safety role

    Legg is a cofounder and Chief AGI Scientist at Google DeepMind.

Sources and credits

Photo: Libor.burian. License: CC BY-SA 4.0. Source: commons.wikimedia.org. Cropped to a square and resized. Ask us to remove or replace this photo.

  1. A Collection of Definitions of Intelligence arxiv.org. Checked 10 October 2026.
  2. Universal Intelligence: A Definition of Machine Intelligence arxiv.org. Checked 10 October 2026.
  3. DeepMind enters its tenth year deepmind.google. Checked 10 October 2026.
  4. Human-level control through deep reinforcement learning nature.com. Checked 10 October 2026.
  5. Deep Reinforcement Learning from Human Preferences arxiv.org. Checked 10 October 2026.
  6. Levels of AGI arxiv.org. Checked 10 October 2026.
  7. Taking a responsible path to AGI deepmind.google. Checked 10 October 2026.
  8. Google DeepMind: responsibility and safety deepmind.google. Checked 10 October 2026.