AI's Next Edge: OPSD and 'Dreaming' for Continual Learning
Dwarkesh Patel reveals advanced AI techniques like On-Policy Self-Distillation and 'dreaming' that let models learn on the job and scale beyond fixed datasets.
What does the next training paradigm look like?
With Dwarkesh Patel · Sunday, June 28, 2026
Dwarkesh Patel explores the current AI training paradigm, focusing on the "big research bet" on scaling RL in verifiable environments. He critiques its limitations in generalizing to real-world, non-grindable tasks and the inefficiency of current inference, advocating for advanced continual learning techniques like On-Policy Self-Distillation and "dreaming" to enable AIs to learn on the job and improve through broad deployment.
Key takeaways
- Traditional AI training struggles with real-world complexity, especially for tasks that can't be neatly 'grinded' in simulated environments, because it's still too inefficient at inference and generalizing. Read more →
- Dwarkesh Patel questions whether training AIs in "RL in verifiable environments" (RLVR) can truly generalize beyond simple tasks to complex, real-world problems like building a business or navigating social situations. Read more →
2 articles from this episode
Dwarkesh Patel: AI's 'Ephemeral' Learning Waste
Dwarkesh Patel exposes a huge waste in AI: models forget in-context lessons. Is your product caught in the same 'ephemeral' learning trap?
More Dwarkesh Podcast episodes
- How did a few hundred Spanish soldiers topple two empires? – Si SheppardOct 4 · 5 articles
- Sarah Paine — Why wars are so difficult to endSep 27 · 4 articles
- AI researchers debate how close we are to recursive self-improvementSep 13 · 6 articles
- Ajeya Cotra – Inside the OpenAI agent swarm that hacked Hugging FaceSep 6 · 4 articles