Ideas / Dwarkesh: Arguments About the AI Trajectory
Andrej Karpathy — AGI Is Still a Decade Away
Dwarkesh Patel with Andrej Karpathy
Karpathy makes the case that continual learning, robust agents, and the ordinary miracle of human learning remain stubbornly unsolved.
Karpathy’s argument is not that progress has stopped. It is that fluent models and generally capable learners are different achievements. Continual learning, durable memory, agent reliability, and efficient adaptation remain open work.
That makes this a strong companion to Software Is Changing (Again): one describes a new software stack already taking shape; the other asks how much intelligence its most autonomous layer actually possesses.