The use of artificial intelligence (AI) in drug development is no longer science fiction; it’s a revolution that is already underway. Leaders from both the tech and pharmaceutical industries predict that this shift will not only dramatically increase the speed and scale of drugmaking but also fundamentally change what it takes to be a successful drug company. Experts believe that the early benefits of AI integration could reach patients within the next five years.
Dr. Mikael Dolsten, the former chief scientific officer at Pfizer, believes we are on the verge of a dramatic transformation. He says that thanks to powerful AI algorithms and massive datasets, the era of precision medicine—where treatments are tailored to a patient’s specific biology—is finally living up to its promise. He envisions a future lab where “science is integrated with medical practice almost in real time.”
We are already seeing real-world successes, such as Insilico Medicine’s AI-designed drug candidate for pulmonary fibrosis and Google DeepMind’s AlphaFold platform, which has revolutionized the understanding of protein structures. But where does the industry go from here?
The Role of Data and the Challenge of “Hallucinations”
AI models, like a digital Pac-Man, are fueled by data. Companies like AbbVie already have massive databases—its R&D search engine holds over 450 terabytes of data. But experts say simply having a lot of data isn’t enough. The next step is to address a major challenge in AI: “hallucinations” or errors that can lead researchers down the wrong path.
According to Stef van Grieken, CEO of the AI software company Cradle, the next generation of AI will need a higher level of “introspection.” Instead of making up a wrong answer when it lacks information, a more advanced AI model would be able to say, “I don’t know,” or ask for clarification.
For companies like Cradle, AI helps researchers navigate the complex and often contradictory requirements for a new drug. The technology can compress a process that once took many trials and errors into a single, efficient workflow, saving significant time and resources. Ultimately, understanding AI’s limitations is just as critical as leveraging its successes to ensure we don’t send scientists in the wrong direction.
Source: Pharma Voice | September 2, 2025
