newsfilter.io

Latest Interviews

Showing 1–3 of 3 interview transcripts.

Clear all filters
  1. Lex Fridman2h 28m

    Demis Hassabis: Future of AI, Simulating Reality, Physics and Video Games | Lex Fridman Podcast #475

    Demis Hassabis, Lex Fridman

    Demis Hassabis presents a comprehensive framework asserting that classical learning algorithms can model natural systems due to evolutionary selection, challenging traditional views on computation and enabling advanced capabilities in video generation and scientific discovery. He forecasts a high probability of achieving Artificial General Intelligence by 2030, driven by AI's ability to solve complex physical problems and accelerate energy solutions like nuclear fusion to create a post-scarcity civilization. While acknowledging existential risks and the need for international safety cooperation, Hassabis envisions a future where AI enhances human flourishing through personalized simulations, open-world games, and the preservation of uniquely human qualities like compassion.

  2. Lex Fridman10 min

    Lex Fridman plays chess with Demis Hassabis

    Lex Fridman, Demis Hassabis

    British Chess Champion Natasha Regan and Matthew Sadler co-authored a book that reveals AlphaZero's development notebooks and introduces its revolutionary chess concepts, including deliberate piece sacrifices for positional control. The publication has been endorsed by International Master Danny Carlson, who has already integrated these strategies into his own gameplay to enhance his competitive edge. Through this collaboration, the authors provide unprecedented access to the hidden methodologies of the AI, bridging the gap between theoretical innovation and practical application for serious players.

  3. Lex Fridman2h 11m

    Demis Hassabis: DeepMind - AI, Superintelligence & the Future of Humanity | Lex Fridman Podcast #299

    Demis Hassabis, Lex Fridman

    Demis Hassabis outlines a comprehensive vision for Artificial Intelligence, detailing how DeepMind has progressed from game-playing benchmarks like AlphaGo to solving the protein folding problem with AlphaFold. He argues that future AI systems must evaluate general intelligence across diverse modalities and physical domains, including nuclear fusion control and material science, while explicitly decoupling intelligence from consciousness to address ethical risks. Ultimately, Hassabis envisions these technologies as foundational tools designed to accelerate the discovery of fundamental laws governing the universe and biology.