Recent stories about AI systems behaving unpredictably have raised serious questions about safety, control and regulation. At the same time, generative AI is already being used in healthcare for medical information, patient communication, administrative tasks and, increasingly, clinical support. Last month’s survey asked two related questions: How concerned are readers about the long-term risks of increasingly powerful AI? And when it comes specifically to clinical medicine, do they believe we’re moving too fast, too slowly or at just the right speed? Here are the results:
My thoughts: I agree with readers that the existential risks posed by increasingly powerful frontier AI models are real and demand serious attention. Nearly three-quarters of respondents (72%) expressed at least some concern that future AI systems could cause serious harm to humanity. To limit the danger, Congress should establish safeguards comparable to those used in aviation, including independent safety reviews and clear requirements for reporting serious incidents. Companies such as OpenAI, Google and Anthropic already conduct safety testing and disclose findings. Federal oversight would establish consistent standards and independent accountability across the industry.
And I also agree with the 39% of readers who say we’re moving too slowly when it comes to clinicians using generative AI applications to improve clinical outcomes and reduce complications for the patients they treat. The opportunity to save tens of thousands of lives annually is too great for medicine to close its eyes to the potential. I’m optimistic that the combination of dedicated clinicians, empowered patients and GenAI can achieve exponentially better outcomes than any of the three alone.
Thanks to all who voted! To participate in future surveys, and for access to timely news and opinion on American healthcare, sign up for my free (and ad-free) newsletter Monthly Musings on American Healthcare.

Doctors should inform patients how/if they are using AI in the treatment setting