Subvolts take Explore AI's future beyond human data with reinforcement learning pioneers. Apr 10, 2025 - Developers
- Learn about alternative AI learning paradigms beyond supervised learning on human data.
- Business
- Understand Google DeepMind's long-term vision for AI progress and potential future capabilities.
This episode features a key leader from Google DeepMind discussing foundational AI research directions.
Is Human Data Enough? With David Silver
Audio streams from Hannah Fry. Your place is remembered on this device.
From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
Google DeepMind's podcast features David Silver discussing the future of AI, contrasting the current "era of human data" with an "era of experience." He uses AlphaGo and AlphaZero as examples of AI learning through reinforcement learning without human knowledge, arguing this path is crucial for advancing AI and achieving superintelligence. The episode explores this approach versus data-dependent large language models.
This episode discusses two ways AI learns: by studying human knowledge or by teaching itself through trial and error. The speaker argues that letting AI learn by itself, like in games, is a powerful way to make it smarter, potentially leading to superintelligence.
This episode offers insight into Google DeepMind's long-term AI strategy, focusing on reinforcement learning and self-discovery. It's for researchers and AI enthusiasts interested in paths beyond current large language model limitations. Silver's perspective connects to ongoing debates about AI's fundamental learning mechanisms and the pursuit of artificial general intelligence, a core goal for frontier labs.
- HostHannah Fry
- GuestDavid SilverVP of Reinforcement Learning, Google DeepMind
- GuestFan Hui
- Guest: David Silver, VP of Reinforcement Learning
- Show: Google DeepMind: The Podcast
- Length: 50 minutes
- Concepts Discussed: "Era of experience" vs. "era of human data"
- Examples: AlphaGo and AlphaZero
In this episode of Google DeepMind: The Podcast, VP of Reinforcement Learning, David Silver, describes his vision for the future of AI, exploring the concept of the "era of experience" versus the current "era of human data". Using AlphaGo and AlphaZero as examples, he highlights how these systems surpassed human capabilities by engaging in reinforcement learning without prior human knowledge. This approach contrasts with large language models, which depend on human data and feedback. Silver emphasizes the need to explore this path to drive AI progress and achieve artificial superintelligence.
Timestamps
- 00:00 Introduction
- 01:50 Era of experience
- 03:45 AlphaZero
- 10:19 Move 37
- 15:20 Reinforcement learning and human feedback
- 24:30 AlphaProof
- 29:50 Math Olympiads
- 35:00 Experience based methods
- 42:56 Hannah's reflections
- 44:00 Fan Hui joins
___
Thanks to everyone who made this possible, including but not limited to:
- Presenter: Professor Hannah Fry
- Series Producer: Dan Hardoon
- Series Editor: Rami Tzabar
- Commissioner & Producer: Emma Yousif
- Music Composition: Eleni Shaw
- Audio Engineer: Richard Courtice
- Production Manager: Dan Lazard
- Video Director and Editor: Bernardo Resende
- Video Studio Production: Nicholas Duke
- Video Editor: Bilal Merhi
- Audio Engineer: Perry Rogantin
- Camera and Lighting Operator: Robert Messere
- Production Coordination: Zoey Roberts, Sarah Ellen Morton
- Visual Identity and Design: Rob Ashley
- Commissioned by Google DeepMind
Please leave us a review on Spotify or Apple Podcasts if you enjoyed this episode. We always want to hear from our audience whether that's in the form of feedback, new idea or a guest recommendation!
Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
- Reinforcement Learning
- David Silver
- A way for AI to learn by trying things and getting rewards or penalties, like training a pet.
- Artificial Superintelligence
- David Silver
- AI that is much smarter than the smartest humans in almost every field.
44 minCan AI help us better predict the weather?
WithPeter Battaglia
45 minThe mathematics of AI uncertainty
WithZoubin Ghahramani
53 minSame guests, other shows
1 h 08David Silver and Lasse Espeholt of Ineffable Intelligence: A Superlearner for the Age of Experience
WithDavid Silver, Lasse Espeholt
27 minLife is like a game
WithOriol Vinyals, Max Jaderberg, Raia Hadsell, David Silver
26 minGo to Zero
WithDavid Silver, Matthew Sadler, Natasha Regan, Demis Hassabis
Questions people ask
- What is the "era of experience" vs. the "era of human data"?
- The "era of human data" relies on AI learning from information created by humans, like current large language models. The "era of experience" involves AI learning through self-play and reinforcement learning, similar to AlphaZero.
- How did AlphaGo and AlphaZero learn?
- AlphaGo and AlphaZero learned through reinforcement learning, playing against themselves and improving without prior human game knowledge, demonstrating the power of the "era of experience."
Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Hannah Fry.