Fork the consciousness, or download the project and create your own.

Shane Legg

Google DeepMind

Functionalist What matters is the organisation of the processing, not the material

Position on AI consciousness

Legg's position is that capability claims need formal definitions, and the 2026 framework he co-authored applies that standard to consciousness, producing credences over functional levels instead of verbal verdicts.

Shane Legg co-founded Google DeepMind with Demis Hassabis and Mustafa Suleyman, and his work on the mathematical definition of intelligence with Marcus Hutter supplied the field’s most cited formal measure, intelligence as expected reward maximisation across environments. His AGI forecasting work, including the long-running probability estimates for human-level machine intelligence, established the practice of stating capability timelines as explicit credences rather than predictions. That history makes him a natural participant in the 2026 consciousness framework.

From Cacophony to Hierarchy, co-authored with thirteen colleagues across Google DeepMind, Oxford and Sussex, extends credence-based thinking from capability to consciousness. The framework places rival theories at levels of functional description, takes indicator evidence with likelihood ratios, and returns a probability that shifts with the user’s stated theoretical weights. Legg’s involvement signals that capability-side researchers are engaging consciousness assessment with the same formality they apply to capability claims. The framework’s supervenience commitment places it in the functionalist family, with substrate-dependent positions carried as explicit realisability constraints.

Known for. Co-founder of Google DeepMind, mathematical definitions of intelligence with Marcus Hutter, AGI forecasting, co-author of the cacophony assessment framework

Coverage on this site

Related researchers