Descriptions:
Two Minute Papers host Károly Zsolnai-Fehér sits down with Demis Hassabis, CEO of Google DeepMind, for a wide-ranging conversation spanning Gemini’s real-world medical applications, the co-scientist research assistant, AlphaFold’s ongoing scientific impact, and the long-term vision for AI in drug discovery.
Hassabis describes co-scientist as a specialized version of Gemini augmented with tools and harnesses for hypothesis generation, literature summarization, and data analysis—essentially an early-stage AI research assistant. On AlphaFold, he notes the system is now used by over 3 million researchers worldwide, and discusses John Jumper’s prediction that someone will eventually win a Nobel Prize by building on AlphaFold-derived techniques—what Hassabis calls a “second-order Nobel.” He considers this plausible given the volume and importance of current AlphaFold-enabled research.
The drug discovery section is the most technically detailed, with Hassabis walking through the full prediction pipeline challenge: binding affinity, ADME properties (absorption, distribution, metabolism, excretion), toxicity, synthesis routes, and biochemical modeling of protein pocket interactions. He argues that as the first wave of AI-designed drugs moves through the regulatory process, the accumulated evidence will enable a reexamination of FDA approval timelines—potentially analogous to how the COVID emergency accelerated mRNA vaccine authorization. Hassabis also discusses his personal use of Gemini for brainstorming and literature review, and shares his view that curing all disease within a decade is a realistic long-term aspiration rather than idle speculation.
📺 Source: Two Minute Papers · Published May 25, 2026
🏷️ Format: Interview
