Google DeepMind CEO Demis Hassabis has laid out a two-step roadmap for AI to overhaul drug development: build the drug design platform first, then convince regulators that AI predictions are reliable enough to compress the clinical trial process that currently takes a decade or more.
Hassabis described the effort in a conversation with Harry Stebbings on the 20VC podcast, "Hvylya" reports. His spinoff company, Isomorphic Labs, is tackling the first phase - solving the chemistry of compound design, toxicity screening, and drug safety properties that follow after protein structure prediction.
"We're focusing on solving the rest of the drug discovery process, which is a lot of chemistry - designing the compounds, checking it's not toxic and all the different properties you need for drugs to be safe," Hassabis said. He expects the full drug design engine to be operational within five to 10 years.
The harder challenge, he acknowledged, is the regulatory bottleneck. Clinical trials still consume years, even when a promising compound is identified. Hassabis said AI can help by simulating parts of human metabolism and stratifying patients to match them with the right drug for their genomic profile. But the real shift, he argued, requires a track record. "The real revolution will come when a few - maybe a dozen or so - AI drugs get through the whole process, and then the government and the regulatory bodies see that," he said.
Once regulators have enough data to back-test model predictions, Hassabis envisions a future where some steps - like animal testing - could be eliminated entirely, and dosage trials could be accelerated. He said the platform Isomorphic Labs is building is general-purpose, designed to work across therapeutic areas including cancer, neurodegeneration, cardiovascular disease, and immunology. "I want to literally cure cancer," Hassabis said. "I know people say that's the cliche, but actually what we're building at Isomorphic is general purpose."
Also read: how one dog's tumor shrank after its owner used AI to design a personalized cancer treatment.