Drugmakers use AI to speed up early research, but trials still take time

In short: Drug companies are using AI more widely to find and design new medicines faster, but the real world testing still takes years.

What's going on

Pharmaceutical companies have used computer-based tools for years, but interest has surged since ChatGPT brought attention to a newer kind of AI. Big drugmakers and smaller start-ups are now spending billions of dollars to use AI in drug research and development.

One key moment was AlphaFold2, released by Google DeepMind in 2021. AlphaFold predicts the 3D shape of a protein from its basic building blocks. You can think of it like folding a paper model from written instructions, except the “paper” is a protein inside the body. Knowing a protein’s shape helps scientists understand how diseases work and where a medicine could attach.

Consultancy Capgemini estimates that AI-driven drug discovery platforms could be involved in about 60 percent of new molecular entities (new drug compounds) within the next decade, up from about 12 percent of today’s drug pipelines. Capgemini’s Thorsten Rall says AI could cut the time to reach pre-clinical testing from four to five years to 12 to 18 months. Companies also hope AI can reduce costs and help recruit patients for clinical trials more efficiently.

What to watch

Experts caution that even if AI improves early research, it cannot “compress the physical world.” Rebecca Paul of Isomorphic Labs says biology still needs time to play out in patients, and regulators still require proof of safety and benefit. Only about one in 10 drugs that enter clinical trials make it to approval, so the big test will be whether AI can lower that failure rate, not just speed up the first steps.

Source: Financial Times