DeepMind CEO says AI speeds drug discovery

Demis Hassabis, CEO of DeepMind and Isomorphic Labs, said AI could reduce drug discovery timelines from years to months.

Isomorphic Labs, an Alphabet unit founded to commercialize AlphaFold, has partnerships with Eli Lilly and Novartis. Hassabis and DeepMind scientist John Jumper won the 2024 Nobel Prize in chemistry for their work on AlphaFold.

The company is now developing an advanced version to better understand complex molecular interactions.

Isomorphic Labs is working on treatments for cancer and immune disorders. Its partnership with Novartis has expanded to six drug targets, up from three last year.

Despite progress, no AI-designed drugs have yet completed clinical trials or reached patients.

Company executives have not provided a timeline for when that might happen.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

AI drug discovery shows early promise but lacks breakthrough clinical proof

While Hassabis promises to cut drug discovery timelines from years to months, no AI-designed drugs have completed successful clinical trials yet, creating a gap between ambitious predictions and proven results1.
However, early indicators suggest progress: AI-discovered drugs show an 80-90% success rate in Phase 1 trials compared to the traditional 40-65%, according to Nature Biotechnology2.
The first concrete milestone came with Rentosertib, developed by Insilico Medicine, which became the first drug where both the target and compound were discovered using AI3.
Insilico reduced their development timeline from target identification to preclinical candidate selection to just 18 months, demonstrating tangible efficiency gains3.
The field has momentum with over 150 small-molecule drugs currently in development using AI techniques, and the FDA granted its first Orphan Drug Designation to an AI-discovered treatment in 20234.

Historical AI cycles suggest caution about timeline predictions

AI in medicine has experienced several “winters” since the 1970s due to unrealistic expectations and implementation challenges, according to historical analysis in medical AI development5.
Hassabis’s January promise that Isomorphic Labs would begin clinical trials by end of year hasn’t materialized, with the CEO now saying “it’s a bit early to say” when asked for updates1.
The pattern mirrors past AI hype cycles where ambitious timelines were followed by periods of reduced funding and interest when expectations weren’t met5.