Summary
Dario Amodei and Demis Hassabis debated how soon AI systems could reach human-level breadth, with coding automation and AI research loops as the likely accelerants. They balanced optimism about scientific breakthroughs with warnings on autonomous control, misuse, authoritarian deployment, and labor disruption.
Key Topics
5 key topics from Demis Hassabis, Dario Amodei and Zanny Minton Beddoes at World Economic Forum Annual Meeting 2026. Thicker branches were mentioned more often in the talk.
The Day After AGI
Click a topic to open it
Key Takeaways
- Track coding automation as the clearest AGI acceleration signal.
- Test AI outputs carefully in domains without fast verification loops.
- Prepare governance plans before autonomous systems close the improvement loop.
- Balance breakthrough optimism with concrete misuse and labor-risk mitigations.
- Build research-led AI organizations around hard scientific problems.
“How do we keep these systems under control that are highly autonomous and smarter than any human?”
Share
Up Next
Next in agenda
Conversation with Satya Nadella, CEO of Microsoft
Satya Nadella, Laurence D. Flink
ai diffusionplatform shift
Also on ai safety
Next Phase of Intelligence
Nicholas Thompson , Eric Xing, Yoshua Bengio, Yuval Noah Harari, Yejin Choi
ai safetycontinual learning
Later in the agenda
What it Takes to Build
Steven Barlett, Jessica Lessin, Bret Taylor
ai agentscreator economy