Demis Hassabis, CEO of Google DeepMind and creator of AlphaFold (the one that got him the Nobel), and Sebastian Mallaby, British journalist and author of the VC bestseller *The Power Law*, took the stage together in San Francisco yesterday.
Here are 9 key quotes from their conversation:
1. "There's a 50% chance OpenAI goes bankrupt in the next 18 months"
— Sebastian Mallaby
Mallaby underscored the extreme financial pressure on leading AI labs from massive compute and energy costs.
2. "Dario is the person I know best among the heads of other labs"
— Demis Hassabis
Dario Amodei is CEO of Anthropic; Hassabis noted the close personal ties between the leaders of the major AI companies.
3. "It's an unsustainable situation when a private company decides who gets access. What happens when China can do the same thing in 6–12 months?"
— Sebastian Mallaby
This was about frontier models being used in cyber defense, and the geopolitical risk of China rapidly reaching parity.
4. "Not every country is pessimistic about AI. I just got back from a summit with Modi in India — the mood there is very optimistic"
— Demis Hassabis
Hassabis contrasted Western hand-wringing over AI risk with the strong enthusiasm in emerging economies like India.
5. "The most exciting work right now is happening at Isomorphic Labs. AlphaFold was just one problem. We need six more 'AlphaFold moments' to compress drug development from 10 years to a few months"
— Demis Hassabis
Isomorphic Labs is DeepMind's biotech arm; Hassabis expects several more AlphaFold-scale breakthroughs that will radically accelerate drug discovery.
6. "I don't throw around probabilities. I just know it's not zero. And when Marc Andreessen and Yann LeCun say 0% — that's insane, in my view"
— Demis Hassabis
p(doom)) is the estimated probability that AI leads to human extinction; Hassabis pushed back on the absolute optimism of some prominent AI figures.
7. "A post-scarcity world. On one side — an explosion of science. On the other — figuring out how to distribute the gains fairly. Plus enormous philosophical questions. We're going to need new great philosophers"
— Demis Hassabis
He laid out the long-term consequences of AGI: unprecedented scientific progress paired with hard questions about distribution and ethics.
8. "Dive into frontier AI tools. Everyone has access to models that are only 3–6 months behind the cutting edge. Massive opportunity — apply AI in areas nobody has touched yet"
— Demis Hassabis
Career advice for developers and practitioners: the gap between frontier and publicly available models is small enough to build real, meaningful innovations.
9. "When I started, I imagined it differently — more like CERN: open exchange of ideas, collaboration, stress-testing. My goal now is to make AI development more scientific, rigorous, and deliberate. We need social scientists and economists at the table. The decisions of the next 5–10 years will shape the next thousand. But I remain very optimistic"
— Demis Hassabis
Hassabis called for a more open, interdisciplinary approach to AI research instead of the closed, competitive lab culture that dominates today.
Full thread with the original recording here 👇 https://x.com/deedydas/status/2046443025975652484