About Demis Hassabis
Demis Hassabis is the CEO and co-founder of Google DeepMind, formed from the 2023 merger of DeepMind and Google Brain. He received the 2024 Nobel Prize in Chemistry for AlphaFold's breakthrough in protein structure prediction.
Career Highlights
- Google DeepMind (2023-present): CEO after merger of DeepMind and Google Brain
- DeepMind (2010-2023): Co-founder and CEO, acquired by Google in 2014
- Nobel Prize in Chemistry (2024): For AlphaFold protein structure prediction
- AlphaGo (2016): First AI to defeat world champion Go player
- Bullfrog/Lionhead: Game designer (Theme Park, Black & White)
- Chess prodigy: Child chess master, second-highest rated under-14 in the world
Notable Positions
On Jagged Intelligence
Hassabis frames the core AGI barrier:
"You would expect from an AGI system that it would be consistent across the board. Current models can win gold medals at the International Math Olympiad while failing basic logic puzzles."
On World Models
He believes language alone can't get us to AGI:
"Language is richer than we thought, but spatial dynamics, intuitive physics, and sensorimotor experience can't be captured in text."
On Scaling
"There's a lot of room between exponential and asymptotic. We operate on 50% scaling, 50% innovation. Both are required for AGI."
Key Quotes
- "Jagged intelligence is the core AGI barrier."
- "World models are essential for embodied AI."
- "Models don't know what they don't know - that's why they hallucinate."
Related Reading
- World Models - Hassabis's path to AGI
- Jagged Intelligence - The consistency problem he identifies
Video Appearances
Jagged intelligence problem
Current AI has 'jagged intelligence' - it can win gold medals at Math Olympiad while failing basic logic puzzles. You would expect from an AGI system that it would be consistent across the board.
at 00:08:00
World models for AGI
On world models: 'Language is richer than we thought, but spatial dynamics, intuitive physics, and sensorimotor experience can't be captured in text.'
at 00:15:00
Genie + Simma training loop
Genie and Simma - 'The two AIs are kind of interacting in the minds of each other' - creating infinite training environments for embodied AI.
at 00:22:00
AGI missing ingredients
At Davos 2026: 'I think there are some missing capabilities at the moment... not just solving existing conjectures but actually coming up with the question in the first place or the theory. That's the highest level of scientific creativity.'
at 00:06:30
Gemini 3 progress
On Google DeepMind's resurgence: 'You can start seeing the progress that's been made in both the models with Gemini 3 but also on the product side with Gemini app getting increasing market share.'
at 00:12:00
AI career advice
On advice for students: 'I would be telling them to get really unbelievably proficient with these tools. There's almost a capability overhang even in today's models.'
at 00:24:00