Andrew Brookes/Getty Images
- Prescription drugs designed by AI could be in clinical trials within years, DeepMind's CEO predicted.
- Demis Hassabis said in an interview he'd be "disappointed" if it took more than a couple of years.
- Clinical trials help evaluate if a drug is safe and effective; drugs may still never hit the market even if they undergo clinical testing.
One of the foremost figures in AI thinks prescription drugs designed by AI could reach clinical testing in a few years.
Google DeepMind CEO Demis Hassabis discussed his prediction on an episode of The New York Times podcast "Hard Fork" released Friday.
"I think we are very close," Hassabis said when asked about whether AI was close to being capable of helping cure a major disease like Alzheimer's or a cancer. "I would say we're a couple of years away from having the first truly AI-designed drugs for a major disease, cardiovascular, cancer."
"We've just signed big deals with Big Pharma and on real drug programs. And I expect in the next couple of years, we'll have AI-designed drugs in the clinic, in clinical testing," he added. "And that's going to be an amazing time. And that's when people will start to really feel the benefits in their daily lives in really material and incredible ways."
Clinical trials help assess if a new drug or device is safe and effective for human use. Drugs that make it to the clinical testing phase of development can still take years to reach the market, if they do so at all.
While some experts say AI can help accelerate the drug discovery process and offer patients personalized care, they also warn of the risk of bias if these models aren't trained on diverse data sets.
AI doesn't always get things right, as Google knows. Hassabis recently spoke of Google DeepMind's fiasco with its Gemini AI image generator. After users pointed out it created historically inaccurate images, including racially diverse images of the Founding Fathers and Nazis, Google paused Gemini's ability to generate images of people last week.
"That, which is a well-intended feature, was applied, it turns out, too bluntly," Hassabis said of the image generator while speaking at the Mobile World Conference in Barcelona. "We care of course about historical accuracy and so we've taken that feature offline while we fix that, and we hope to have that back online in very short order."