AI in Oncology: Hype, Evidence, and the Threshold of Clinical Trust

What does it take for AI to move from technical innovation to real-world adoption in oncology?

In the first episode of Precision Signals, Dr. Sean Khozin speaks with Dr. Cora Sternberg and Dr. Olivier Elemento—two leaders at the intersection of clinical trials, precision medicine, and computational biology.

Together, they unpack the central paradox of medical AI: why so many tools are technically validated, yet so few are clinically trusted. Topics include:

• Why randomized clinical trials—not just retrospective benchmarks—are essential for AI adoption
• The limitations of current training data and the risks of treating clinical guidelines as ground truth
• The evolving role of real-world data, EHRs, and multimodal inputs in building more adaptive AI systems
• What physicians actually need to trust and use AI in everyday cancer care

This is a candid conversation about evidence, trust, and what it will take to build AI that earns a place in the oncology clinic, grounded in insights from the participants’ recent review published in NEJM AI: https://ai.nejm.org/doi/abs/10.1056/A...