TL;DR: AI isn’t just helping scientists discover drugs—it’s imagining them from scratch, cutting years off development, and opening the door to treatments we never thought possible.
Tucked away in a modest lab in Shanghai, scientists at Insilico Medicine did something remarkable: they created a new drug candidate in just 18 months. Not by accident. Not by trial and error. But with the help of artificial intelligence (AI). From identifying the right protein to targeting the disease and shaping the molecule itself, AI guided the entire process.
It might sound like something pulled from the pages of science fiction, but it’s happening right now. This is the world of generative AI in drug discovery, and it’s changing the rules of the game.
The digital chemist with a wild imagination
Think about traditional AI as a super-sleuth. It’s amazing at scanning mountains of data and spotting patterns humans might miss. But generative AI? It goes beyond observation. It imagines. Like a digital chemist with an endless supply of creativity, it builds entirely new molecules and sometimes ones that look like nothing seen in nature.
These tools are driven by powerful algorithms, like GANs (generative adversarial networks) and transformers. Platforms like Chemistry42 (from Insilico), AtomNet (from Atomwise), and Exscientia’s Centaur Chemist are making it easier and faster than ever to design promising new drugs.
Meet the movers and shakers
Here are five companies using AI not just as a tool, but as a true co-pilot in the lab:
- Insilico Medicine
Under the guidance of founder Alex Zhavoronkov, Insilico hit a major milestone: getting an AI-designed drug, Rentosertib, into Phase 2 clinical trials for a lung disease called idiopathic pulmonary fibrosis. What’s different here? The AI didn’t just come up with a molecule—it also pinpointed the biological target. “This was not just a molecule,” Zhavoronkov explained, “but the first time AI identified both the biological target and the compound.” ¹
- Atomwise
Atomwise uses its AtomNet platform to screen billions of potential compounds—virtually. That means they can sort through the noise and zero in on promising candidates fast. By the end of 2023, they had chosen a TYK2 inhibitor for clinical development.². As co-founder Abraham Heifets put it, “To discover something new, you have to start in new places.”
- Exscientia
In just 12 months, Exscientia went from designing a new compound for OCD to launching clinical trials. ³,⁶ That’s a fraction of the time drug development usually takes. CEO Andrew Hopkins says it best: “AI doesn’t replace chemists—it helps them work smarter and faster.”
- BenevolentAI
When COVID-19 first hit, scientists were racing against the clock. BenevolentAI used its system to find a new use for an old arthritis drug—baricitinib—which turned out to have both antiviral and anti-inflammatory potential. It took their AI just three days to flag the candidate. ⁴,⁷ Clinical studies later backed it up.
- Recursion Pharmaceuticals
Recursion mixes biology with high-speed image analysis and machine learning to explore new targets. One of their AI-assisted wins? A cancer drug that goes after the RBM39 protein was discovered and advanced in under 18 months. ⁵ Their team sees it as a sign of what’s possible when biology and data science work together.
Fast, but not without challenges
The results are exciting. Drug timelines that used to take up to a decade can now be cut down to a couple of years—or even months. Some companies are seeing success rates far beyond industry norms. The savings in both time and cost are enormous.
But this is still a work in progress. These AI systems need massive, high-quality datasets to function well. And no matter how smart the algorithm is, humans still need to validate and refine what the models suggest. On top of that, regulatory agencies are trying to keep up with the rapid pace of innovation.
Still, momentum is building. Pharmaceutical giants are investing heavily in AI startups. Researchers are getting more comfortable with data-driven discovery. And regulatory agencies are beginning to adapt.
Where this is all going
Generative AI isn’t about replacing scientists. It’s about giving them superpowers. Imagine taking years of lab work and distilling it into a few weeks of high-powered simulations. That’s the promise here.
It’s also changing how we think about medicine. Instead of reacting to disease, we’re starting to design solutions from scratch—with intention and speed.
As Andrew Hopkins predicted, “In the future, every new drug will be born with AI.”
And that future is arriving fast.
References
- Zhavoronkov A, Aliper A, Wang J, et al. Deep learning enables rapid identification of potent DDR1 kinase inhibitors. Nat Biotechnol. 2021;39(8):1034-1042.
- Atomwise. AtomNet technology platform. https://www.atomwise.com/technology/. Accessed June 5, 2025.
- Hopkins AL, Groom CR, Alex A. Ligand efficiency: a useful metric for lead selection. Drug Discov Today. 2004;9(10):430-431.
- Richardson P, Griffin I, Tucker C, et al. Baricitinib as potential treatment for 2019-nCoV acute respiratory disease. Lancet. 2020;395(10223):e30-e31.
- Recursion Pharmaceuticals. Our pipeline. https://www.recursion.com/pipeline/. Accessed June 5, 2025.
- Exscientia. AI-designed drug reaches clinical trials in record time. https://www.exscientia.ai/news. Published February 2020. Accessed June 5, 2025.
- BenevolentAI. Using AI to fight COVID-19. https://www.benevolent.com/covid19/. Accessed June 5, 2025.
Leave a Reply