Shane Legg co-founded DeepMind in 2010 with Demis Hassabis and Mustafa Suleyman, driven by his academic work on machine super intelligence. His 2008 PhD thesis at IDSIA under Marcus Hutter formalized a mathematical definition of machine intelligence based on algorithmic information theory, one of the earliest rigorous attempts to define AGI. Legg was sounding alarms about existential risk from artificial general intelligence years before it became mainstream discourse. At DeepMind he has served as Chief AGI Scientist, guiding the long-term research agenda toward safe, capable general intelligence. He predicted a 50% chance of human-level AI by 2028, a forecast that increasingly looks prescient. Legg represents the rare combination of someone who both builds frontier AI systems and takes their dangers seriously enough to have co-founded an organization explicitly aimed at solving intelligence safely.
Key contributions
- Co-founded DeepMind alongside Hassabis and Suleyman
- Formalized a universal mathematical definition of machine intelligence
- Early advocate of AGI existential risk, years before mainstream awareness
- Guides DeepMind's long-term AGI safety research agenda