Ilya Sutskever: From the Age of Scaling to Reconstruction

Why it matters

In November 2025, Ilya Sutskever told Dwarkesh Patel we are moving from the age of scaling to the age of reconstruction — on the limits of next-token prediction, what reasoning models really mean, and why AI's future is smarter, not bigger. A defining interview for understanding the 2025 paradigm shift.

Key takeaways

• The 'reconstruction era': after next-token prediction hits a ceiling, AI needs new learning paradigms.
• Reasoning models matter not for longer thinking but for changing how models model the world.
• Superintelligence risk remains real; alignment needs research now.
• His AGI timeline judgment became more cautious after leaving OpenAI.
• Advice to young researchers: find genuinely important problems, not trendy ones.