Headline

DeepMind CEO Demis Hassabis says his teams are using AI to tackle cancer through AlphaFold and Isomorphic Labs.

Summary

When Luiza Jarovsky tweeted calling for AI leaders to join an "AI race to cure cancer," Hassabis replied that his team is already on it. They're using AlphaFold for protein structure prediction and Isomorphic Labs for drug discovery. AlphaFold models biological structures to help researchers understand diseases faster, while Isomorphic Labs works on designing new drugs. This is a concrete example of foundational AI models being applied to actual medical problems rather than demos and benchmarks.

Analysis

AlphaFold, which won a Nobel Prize, predicts protein structures with atomic accuracy and has generated over 200 million entries since 2020. For cancer research, this means faster insights into the molecular interactions behind the disease.

Isomorphic Labs, an Alphabet spin-off from DeepMind, builds on this foundation. Their proprietary IsoDDE engine reportedly outperforms AlphaFold 3 at predicting how proteins interact with potential drug molecules. The company has $3 billion in partnerships with Eli Lilly and Novartis to work on oncology drug pipelines.

What's worth watching: Isomorphic's tools are closed-source, which could give them an edge over open alternatives but also raises questions about who gets access. Clinical trials are expected by late 2026, and that's when we'll actually know if AI can speed up drug development or just generate impressive papers.

The broader picture: more companies are betting on AI for precision medicine, but there's real tension between proprietary and open approaches that nobody has figured out yet.

Impact Assessment

• Significance: High
• Categories: AI Research, Market Impact, Partnership