Radio Advisory 276: The AI gold rush is changing how humans (and clinicians) make decisions
Nov 18, 2025
David Woods is a veteran researcher in patient safety and cognitive systems engineering, while Mike Rayo and Dane Morey are specialists from The Ohio State University's Cognitive Systems Engineering Lab. They delve into the complexities of AI adoption in healthcare. Key topics include how AI can mislead clinicians, the necessity of designing AI as a joint human-machine system, and the risks of premature adoption without rigorous evaluation. They also share insights from recent studies on AI's impact on nurse decision-making, highlighting the potential dangers of overtrusting AI recommendations.
AI Snips
Chapters
Transcript
Episode notes
AI Is Not One Monolithic Tool
- AI is a family of technologies, not a single tool, and large language models can trick people into over-ascribing intelligence.
- Treat AI as fallible support that amplifies human work rather than a standalone solution.
Design For Joint Human–AI Work
- Clinical work is joint activity where people and technology must collaborate to handle variability and surprises.
- Design thinking should evaluate AI inside the human-machine system, not in isolation.
Require AI To Signal When It’s Wrong
- Verify that clinicians can detect when an AI is wrong before deploying it in practice.
- Ensure AI conveys both poor and strong performance clearly to users.