FHC #220: How AI is challenging old assumptions in medicine

Posted on by

Podcast: Play in new window | Download

Subscribe: Spotify | Email | RSS | More

In this Diving Deep episode, Dr. Robert Pearl and Jeremy Corr explore what a breakthrough in mathematics can teach healthcare about generative AI, outdated assumptions and the future of medical practice.

The conversation begins with Paul Erdős, the brilliant Hungarian mathematician whose unit-distance problem stumped experts for 80 years. Generations of mathematicians tried to prove Erdős’ conjecture using the tools and assumptions of geometry. Then OpenAI announced that one of its models had found a different path, one that challenged the assumptions of humans.

For Pearl, the lesson for medicine is not about math. It is about the danger of staying trapped inside old models of thought. American healthcare faces persistent crises in quality, access and affordability, yet the proposed fixes focus on small adjustments to a system that has failed for decades. Pearl argues that generative AI will not reach its full potential if clinicians continue to use it merely for administrative tasks, documentation support or occasional diagnostic assistance.

Instead, he says, medicine must be willing to abandon three longstanding fallacies.

The second half of the episode shifts the discussion from assumptions to urgency. Pearl compares generative AI in medicine to the sudden invention of cars in a world where people could only travel by foot, bicycle or horseback. If faster transportation could save far more lives than it cost, society would not wait until cars were risk-free before building roads and teaching people to drive. He argues medicine should think similarly about GenAI: not as a finished product, but as a rapidly improving tool that could save lives if implemented wisely and quickly.

In the episode, Pearl highlights three areas where GenAI could have immediate impact. In each case, he returns to the same point: the greatest risk is not only that AI might make mistakes, but that medicine will ignore the harms already caused by today’s failures.

For more, tune into this month’s episode and check out the links below.

Helpful links:

* * *

Dr. Robert Pearl is the author of ChatGPT, MD: How AI-Empowered Patients & Doctors Can Take Back Control of American Medicine.

Fixing Healthcare is a co-production of Dr. Robert Pearl and Jeremy Corr. Subscribe to the show via Apple, Spotify or wherever you find podcasts. Join the conversation or suggest a guest by following the show on X and LinkedIn.