Y Combinator Startup Podcast

How to Build the Future: Demis Hassabis

Y Combinator Startup Podcast·May 1, 2026

OVERVIEW

This podcast episode features Demis Hassabis, CEO of Google DeepMind, discussing the past, present, and future of artificial intelligence, particularly focusing on the journey toward Artificial General Intelligence (AGI). He shares insights into DeepMind's foundational work, current challenges in AI development, and the potential impact of advanced AI on scientific discovery and various industries.

KEY TOPICS

  • The path to Artificial General Intelligence (AGI)
  • Unsolved problems in AI like continual learning and long-term reasoning
  • DeepMind's achievements including AlphaGo and AlphaFold
  • The role of distillation in making AI models efficient and accessible
  • Challenges and future directions in scientific discovery using AI
  • The importance of multimodality and agentic systems
  • Strategic considerations for building deep tech companies in the age of AGI
  • The role of open source and ethical considerations in AI development

MAIN TAKEAWAYS

The path to AGI is challenging, with fundamental issues in continual learning, long-term reasoning, and robust memory still largely unsolved. However, existing foundational models and techniques, especially DeepMind's work on agents and reinforcement learning, are crucial building blocks.

DeepMind excels at distilling powerful large models into smaller, highly efficient "flash" models, which is essential for deploying AI at scale across diverse devices for billions of users. This also lowers cost and latency, making AI more accessible.

AI is poised to revolutionize scientific discovery. Tools like AlphaFold demonstrate the power of AI to solve grand challenges in fields like biology, and similar breakthroughs are expected in materials science, mathematics, and medicine. The long-term vision is to use AI to solve fundamental root node problems in science.

Multimodality, the ability of AI to process and integrate different types of data (text, image, audio), is critical for developing AI systems that can understand and interact with the physical world, especially for applications in robotics and digital assistants.

When building a deep tech company today, it's crucial to anticipate the advent of AGI and design systems that can leverage it as a tool rather than being replaced by it. This implies a future where specialized AI systems might be orchestrated by a more general AGI.

Demis emphasizes that truly impactful and long-lasting innovation often comes from tackling hard, deep tech problems, combining AI with other scientific domains through interdisciplinary teams.

NOTABLE QUOTES

"Continual learning, long-term reasoning, some aspects of memory, these are still unsolved. I think all of these are going to be required for AGI."
"My impression is at the moment we're all experimenting a lot of things, but we're only in the maybe the last couple of months starting to find the really valuable places."
"I always thought AI would be the ultimate tool for science."
"The thing I often say that in science is, can it come up with a hypothesis that's really interesting, not just solve one."
"What's the thing that you know now about building at the frontier that you wish you'd known at 25? I think we covered some of it in terms of actually you work out that going after hard problems and deep problems is no more difficult in some ways than going after a shallower, simpler, more superficial problem."

Summarized with DriftNote — AI-powered podcast summaries

Try it free