Google DeepMind: The Podcast Is Human Data Enough? With David Silver
314 snips
Apr 10, 2025 
David Silver, an Original DeepMind researcher renowned for his work on AlphaGo, shares his bold vision for AI's future. He discusses how AI should evolve beyond reliance on human data, advocating for self-learning systems capable of generating their own experiences. Silver reflects on significant moments like move 37 in Go, illustrating AI's potential for creativity. He also introduces AlphaProof, a groundbreaking system for AI-driven mathematical proofs, emphasizing the importance of trust and innovative approaches in the AI landscape.
AI Snips
Chapters
Transcript
Episode notes
Era of Experience
- David Silver introduces the "era of experience" in AI, contrasting it with the current "era of human data".
- He argues that AI needs to move beyond human knowledge to discover new things.
AlphaGo's Breakthrough
- AlphaGo, using no human data (AlphaZero), became the strongest Go program.
- It learned through trial and error, discovering novel strategies like "Move 37".
LLMs Lack of Breakthroughs
- David Silver suggests that large language models (LLMs) haven't had a "Move 37" moment yet.
- He believes this is due to their reliance on human data, limiting their creativity.