Google’s AI boss made OpenAI issue code red. Now he wants to solve disease. | Titans and Disruptors

Key Concepts

• Artificial General Intelligence (AGI): The hypothetical ability of an AI to understand, learn, adapt, and implement knowledge across a wide range of tasks, much like a human.
• DeepMind: An AI research company founded by Demis Hassabis, acquired by Google in 2014, now operating as Google DeepMind.
• Gemini: Google’s latest and most capable AI model, competing with OpenAI’s models.
• AlphaFold: A DeepMind AI system that accurately predicts protein structures, revolutionizing biology and drug discovery.
• Isomorphic: A startup founded by Demis Hassabis focused on using AI to accelerate drug discovery and “solve” disease.
• Agentic AI: AI systems capable of autonomous action and decision-making to achieve specific goals.
• Kaizen: A Japanese philosophy of continual improvement.
• Reinforcement Learning & Deep Learning: Core machine learning techniques used in AlphaGo and other DeepMind breakthroughs.

The AI Revolution: A Conversation with Demis Hassabis

I. The Pivotal Moment: DeepMind’s Acquisition & The Rise of Competition

The conversation centers around Demis Hassabis, CEO of Google DeepMind and founder of Isomorphic, and his perspective on the current state and future of AI. The discussion highlights the significance of Google’s acquisition of DeepMind in January 2014 for $500 million. Hassabis and his team intentionally chose Google over a higher offer from Meta (Facebook) due to Google’s superior computational power and commitment to their mission of solving intelligence and applying it to scientific problems. This acquisition, he believes, was a transformative moment in business history, and directly spurred the creation of OpenAI by Elon Musk and Sam Altman, who feared Google’s potential monopoly in the AI space. The success of AlphaGo in 2016, the first program to defeat a world champion in Go, is identified as a watershed moment that ignited the modern AI era.

II. The Foundations: From Chess & Astronomy to AI

Hassabis’s personal journey is presented as integral to his approach to AI. His childhood fascination with fundamental questions about the universe (astronomy, cosmology, physics) and his strategic thinking honed through chess cultivated a desire to understand intelligence itself. He combined a computer science degree with a PhD in cognitive neuroscience, leading to the founding of DeepMind in 2010 with the ambitious goal of achieving Artificial General Intelligence (AGI). He emphasizes that AI is a tool for understanding both the human mind and the universe.

III. AlphaFold: Solving a 50-Year Grand Challenge

A significant portion of the conversation focuses on AlphaFold, DeepMind’s AI system that solved the protein folding problem – a 50-year challenge in biology. AlphaFold can predict the 3D structure of proteins from their amino acid sequence with remarkable accuracy. This breakthrough is crucial because protein structure dictates function, impacting everything from muscle function to disease development. DeepMind has made the structures of over 200 million proteins freely available in a public database, used daily by over 3 million researchers worldwide. Hassabis received the Nobel Prize in Chemistry and a knighthood in 2024 for this achievement.

IV. Isomorphic: AI-Driven Drug Discovery & “Solving” Disease

Isomorphic, Hassabis’s startup, builds upon the foundation of AlphaFold to accelerate drug discovery. While knowing a protein’s structure is vital, Isomorphic aims to predict how drug compounds will bind to proteins and their effects. The traditional drug development process takes 10-15 years and costs billions of dollars with a low success rate (around 10%). Isomorphic aims to drastically reduce this timeline and cost by performing the majority of the search and design process in silico (using computer simulations), reserving wet lab validation for the final stages. The company has partnerships with pharmaceutical giants like Eli Lilly and Novartis and has 17 active drug programs, focusing initially on cancer, cardiovascular disease, and immunology. Hassabis clarifies that Isomorphic aims to “solve” disease, not just cure it, implying a more comprehensive approach.

V. Google DeepMind: Reclaiming Innovation & The Power of Scale

In 2023, Google merged its DeepMind and Brain AI teams under Hassabis’s leadership to streamline innovation and leverage combined resources. Hassabis describes this as a rebuilding of Google’s AI infrastructure, enabling faster deployment of advanced models like Gemini into existing products (Search, YouTube, Chrome). He emphasizes the importance of a “shipping culture” – rapid iteration, risk-taking, and a focus on delivering impactful results. He credits Google’s founders, Larry Page and Sergey Brin, for their renewed engagement and contributions to AI development. He highlights the importance of a balance between scientific rigor and commercial success, emphasizing the need to fund fundamental research like AlphaFold while also generating revenue to sustain innovation.

VI. The Future of AI: Agents, Glasses, and a New Renaissance

Hassabis predicts that 2026 will be a pivotal year for AI, with the emergence of more capable AI agents capable of autonomous task completion. He anticipates the development of AI-powered smart glasses as a “killer app,” offering a “universal assistant” that seamlessly integrates across devices and aspects of daily life. He envisions a future of “radical abundance” driven by AI-powered breakthroughs in energy, materials science, and space exploration, leading to a “new golden era of discovery” and a “new renaissance.” He stresses the importance of responsible AI development, acknowledging both the potential benefits and risks, and the need for a scientific and ethical approach.

Notable Quotes:

• “We did [know the DeepMind acquisition would be big]. Actually, those of us involved in the science.” – Demis Hassabis
• “What mattered to me was not the money. It was the mission and being able to accelerate our progress towards artificial general intelligence.” – Demis Hassabis
• “I think this is one of the benefits of being as part of Google and alphabet, was having the resources and the time to really go after these sort of deep scientific problems.” – Demis Hassabis
• “We’re going to be in a kind of new golden era of discovery. That’s why I hope a kind of new renaissance.” – Demis Hassabis

Data & Statistics:

• DeepMind Acquisition: $500 million (2014)
• Drug Development Timeline: 10-15 years on average
• Drug Development Success Rate: ~10%
• AlphaFold Database: Over 200 million protein structures predicted and publicly available.
• AlphaFold Usage: Over 3 million researchers use AlphaFold daily.
• Alphabet Stock Performance (2023): +65% (best performance since 2009)

Logical Connections:

The conversation flows logically from the historical context of DeepMind’s founding and acquisition, to the scientific breakthroughs achieved (AlphaFold), to the current state of AI development within Google DeepMind, and finally to a vision of the future powered by AI. Hassabis consistently connects his personal motivations and scientific curiosity to the broader goals of solving fundamental problems and improving human life. The discussion highlights how each advancement builds upon the previous one, creating a momentum towards increasingly powerful and impactful AI systems.