McKinsey urges pharma R&D overhaul to unlock AI's full potential

Big pharmaceutical companies have recently announced multiple AI partnerships—Eli Lilly, Bristol Myers Squibb, and Incyte in May, followed by Alnylam and Merck in June—signaling broad industry interest in applying machine learning to drug development. However, McKinsey's report identifies a critical limitation: simply layering AI onto existing R&D workflows optimizes individual steps without creating the systematic feedback loops needed to compound learning across the entire development pipeline. The consultancy's closed-loop model addresses this by repositioning R&D as a connected feedback system rather than a series of isolated gates, where outputs from one decision point become inputs to inform the next.

The potential payoff is substantial: clinical development remains the highest-cost and longest-duration phase of drug development, and McKinsey argues that closed-loop AI can compress both trial duration and decision cycles. Real-world examples such as Robin—which autonomously generated hypotheses, proposed and ran experiments, and interpreted results to guide follow-up work—demonstrate that this is not theoretical. Some companies, particularly techbio startups like Recursion Pharmaceuticals, already characterize their platforms as closed-loop systems combining phenomics, omics layers, AI-driven chemistry design, and clinical intelligence. McKinsey's recommendation is pragmatic: companies should build a blueprint for their desired system across all five decision points, then use it to identify early wins and phase investment over time, avoiding an all-or-nothing approach.

No comments yet. Be the first to share your thoughts!