Can AI Design Better Clinical Trials Than Humans?

Can AI actually become better at designing trials than the industry itself?

In juxtaposition to this question is the fact that, at this point, only a handful of AI-discovered or AI-designed drugs may have reached human trials, and none have yet made it to clinical approval. To be fair, considering that it takes, on average, up to 11 years to develop a new drug (DiMasi et al., 2016), we are still at the very beginning. Not enough time has passed quite yet to draw meaningful conclusions here. But check out this recent example led by Stanford professor James Zou: a “virtual biotech” comprising up to 37,000 AI agents, coordinated by an AI “chief scientist”, identified a promising lung-cancer treatment strategy and demonstrated how AI can rapidly generate evidence.

We are truly at a pivotal point. Exciting times that open the door to much bigger conversation, debate and new regulatory frameworks.

AI in clinical trials is our bread and butter. Undoubtedly. This is what we develop. We are an impact-driven organization, and our mission is to dramatically reduce the time and cost of clinical trials. We applaud any progress in that direction. But we also carefully evaluate and validate it. On a broader societal scale, we are observing these developments with curiosity while building the first end-to-end solution for clinical trial development that leverages AI and technology to simulate trial design, with the goal of reducing failure and cost.