Google DeepMind's Demis Hassabis: How AI could become dangerous (with Reid Hoffman) | Summit '26

November 5, 2026 | 25:32

• 00:00Introducing Demis Hassabis
• 2:54How AlphaFold cracked biology’s 50-year grand challenge
• 5:37Recursive self-improvement and the AlphaZero lesson
• 8:47The danger of closed-loop RSI
• 12:52Industry standards, auditors and working together across borders
• 15:32Can AI be conscious?
• 17:32Why to build intelligent tools, not conscious entities
• 22:08AI sovereignty for companies and countries