Finding New Drugs Is Laborious. This AI Startup Is Skipping The Lab Work
wired ·
Potential Sciences, a UK-based startup backed by the government’s SovAI fund, aims to accelerate drug discovery by using AI models trained on simulated biochemical data rather than traditional lab experiments. The company plans to generate large-scale synthetic datasets based on physics principles to predict how molecules interact with human biology, thereby identifying promising drug candidates faster. This approach seeks to bypass the costly and time-consuming process of physical experimentation, potentially making it economical to develop treatments for rare diseases. While industry experts acknowledge the potential for revolutionizing drug development, they caution that current protein engineering capabilities are still rudimentary compared to historical benchmarks. Despite significant investment and access to national supercomputing resources, Potential Sciences faces intense competition in the crowded AI drug discovery market. Furthermore, the startup must navigate challenges regarding revenue generation, data availability, and the dual-use risk of its technology being misused for harmful purposes. Ultimately, while AI may speed up initial candidate identification, the final stages of safety and efficacy testing will still require substantial time and resources.
wired