self-play
Type: Concept
As of November 2026, self-play has been mentioned in 2 episodes across 2 podcasts, first heard Jul 2025 on Wes and Dylan.
Self-play is a training technique where AI agents learn by competing or cooperating with copies of themselves, often used in reinforcement learning.
| Mentioned in | 2 episodes on 2 podcasts (as of November 2026) |
|---|---|
| First heard | Jul 2025 on Wes and Dylan |
| Field | Machine Learning |
How do different shows talk about self-play?
2 shows discussed self-play in these episodes.
- Dwarkesh Patel — 1 episode
- Wes and Dylan — 1 episode
Who talks about self-play on podcasts?
Dwarkesh Patel brought up self-play; the most recent mention was Nov 2025 on Dwarkesh Patel.
What is self-play mentioned alongside?
self-play is most often mentioned alongside generalization, Silicon Valley and Yan LeCun.
Mentioned together: generalization, Silicon Valley, Yan LeCun, Demis Hassabis, Gemini, scaling laws, Giacomo Rizzolatti, Sam McCandlish, DeepSeek R1, Anthropic