In brief
Health AI is arriving faster than the rules for it, so the answer is neither a ban nor a free-for-all. Laura Adams, Vardit Ravitsky and Roxana Daneshjou argue for shared principles, new ways of regulating, honest disclosure to patients, and routine testing for bias. They treat AI as a tool to deploy responsibly now, because its benefits are real and so are its risks.
The central tension is speed against safety. Adams fears old-style regulation will stifle innovation without preventing harm. Ravitsky asks whether AI is held to a higher standard than error-prone human doctors, while still insisting patients be told when AI is used. Daneshjou shows bias comes from data and from humans, so fixing the model alone is not enough.
Questions
What is the National Academy of Medicine's AI Code of Conduct?
It is a set of principles and six short commitments, led by Laura Adams, meant as a touchstone for organizations writing their own AI rules. It was drafted last year and put out for public comment, and a revised version is coming.
Should patients be told when AI writes their doctor's messages or takes notes?
Vardit Ravitsky says yes, at least for now. In her survey patients liked AI replies until they learned who wrote them, and she argues informed consent and long-term trust require disclosure while AI use is not yet the norm.
Do AI chatbots show bias in medical answers?
Yes. Roxana Daneshjou's team tested several models, including GPT-3.5, GPT-4 and Claude, and all gave some race-based medical answers. One example was advising race be used to calculate kidney function, which she calls a false trope.
Will AI replace doctors like dermatologists or radiologists?
Daneshjou doesn't think it is close for dermatology. She says the job involves skin exams, biopsies, judging intermediate cases and choosing treatments, and AI works best on well-defined tasks, which much of medicine is not.
Why is regulating generative AI in healthcare so hard?
Laura Adams says generative AI is "wickedly challenging" to regulate and that old methods haven't even kept up with earlier AI. She worries they would stifle innovation without stopping harm, so she favors new approaches like regulatory sandboxes (safe testing zones with regulator oversight).
Facts
Stated by the speakers. Each quote matches the transcript, which shows what was said, not that it is true.
Peter Lee, Carey Goldberg and Zak Kohane published the book The AI Revolution in Medicine shortly after OpenAI released GPT-4.
“Shortly after OpenAI's GPT-4 was publicly released, Kerry Goldberg, Dr. Zak Kohani, and I published The AI Revolution in Medicine”
0:00:38 · Peter Lee · Book publication · 2023
OpenAI's ChatGPT launched in November 2022 and GPT-4 followed in the spring of 2023, according to Peter Lee.
“So, November 2022, CHAT GPT. Later in the spring of 2023, GPT-4.”
0:08:12 · Peter Lee · ChatGPT timeline · 2022
Laura Adams says the National Academy of Medicine was formed more than 50 years ago by the federal government as an independent advisory body.
“National Academy was formed more than 50 years ago. It was formed by the federal government, but it is not the federal government.”
0:04:59 · Laura Adams · NAM history
The National Academy of Medicine was formerly the Institute of Medicine, which produced the publications To Err Is Human and Crossing the Quality Chasm.
“some probably know of the National Academy of Medicine when it was Institute of Medicine and produced such publications as to Air is Human and Crossing the Quality Chasm”
0:04:59 · Laura Adams · NAM publications
The National Academy of Medicine's AI Code of Conduct includes six commitments based on complex adaptive systems and simple rules.
“especially the six commitments that were based on the idea of complex adaptive systems, simple rules”
0:06:10 · Laura Adams · AI Code of Conduct
In building the AI Code of Conduct, the team reviewed at least 60 existing sets of AI principles, guidelines and frameworks.
“We looked at least at 60 different sets of these principles, guidelines, and frameworks.”
0:18:09 · Laura Adams · Landscape review
Laura Adams says the review of existing AI principles found environmental impact was a significant gap.
“Also, the impact on the environment was a significant gap as well.”
0:18:45 · Laura Adams · AI principle gaps
The National Academy of Medicine opened its draft AI Code of Conduct principles for public comment after drafting them the prior year.
“we opened it up for public comment once we did the draft last year”
0:19:15 · Laura Adams · Public comment
Peter Lee says Laura Adams led the Rhode Island Quality Institute, which he regards as an important early example of using health data to improve health statewide.
“Also, our listeners might want to know that you led something called the Rhode Island Quality Institute, which was really, I think, arguably the most important early kind of examples”
0:16:59 · Peter Lee · Rhode Island Quality Institute
An article by Bakul Patel and David Blumenthal proposed regulating generative AI somewhat like a licensed and qualified healthcare provider, according to Laura Adams.
“by Bakul Patel and David Blumenthal, who posited, should we be regulating generative AI as we do a licensed and qualified provider?”
0:26:15 · Laura Adams · AI regulation
Laura Adams chairs the Global Opportunities Group for a regulatory and innovation AI sandbox in the United Kingdom.
“I chair the Global Opportunities Group for a regulatory and innovation AI sandbox in the UK.”
0:26:51 · Laura Adams · AI sandbox
Vardit Ravitsky says the Hastings Center has existed for more than 55 years and is considered one of the organizations that founded the field of bioethics.
“We've been around for more than 55 years, and we're considered one of the organisations that actually founded the field known today as bioethics”
0:32:26 · Vardit Ravitsky · Hastings Center
A survey of hundreds of respondents by Vardit Ravitsky and co-authors found a statistically significant difference in satisfaction between human-written and AI-written messages from clinicians.
“We surveyed hundreds of respondents. And basically what we discovered is that there was a statistically significant difference in their level of satisfaction”
0:48:45 · Vardit Ravitsky · Patient message survey
In Vardit Ravitsky's survey, respondents preferred AI-written clinician messages, which were longer, more detailed and conveyed more empathy.
“it turns out that they preferred the messages written by AI. They were longer, more detailed, even conveyed more empathy.”
0:49:34 · Vardit Ravitsky · AI message preference
In Vardit Ravitsky's survey, patients were less satisfied with messages once they were told that AI had written them.
“But then when we disclosed who wrote the message, they were less satisfied when they were told it was AI.”
0:49:34 · Vardit Ravitsky · AI disclosure
Roxana Daneshjou says the majority of large genetic studies were conducted on people of European ancestry, leaving many global populations excluded.
“the majority of these genetic studies were on European ancestry individuals”
1:03:53 · Roxana Daneshjou · Genetics bias
A Science Advances paper from Roxana Daneshjou's team built a diverse skin-disease benchmark and found AI models performed significantly worse on brown and black skin.
“and one of our landmark papers, which was in Science Advances, showed we created a diverse data set, our own diverse benchmark of skin disease and showed that the models performed significantly worse on brown and black skin”
1:05:48 · Roxana Daneshjou · Skin tone bias
Daneshjou's team found that fine-tuning dermatology AI models on diverse skin tones eliminated the performance bias.
“because when we fine-tuned on diverse skin tones, you could make that bias go away.”
1:05:48 · Roxana Daneshjou · Bias mitigation
A Proceedings of the National Academy of Sciences paper on untrue race-based beliefs among medical trainees inspired Daneshjou's team to test whether language models repeat them.
“there was a paper in the Proceedings of the National Academy of Sciences, which did not look at technology at all, but actually looked at the race-based biases of medical trainees”
1:09:12 · Roxana Daneshjou · Race-based medicine
Daneshjou says GPT-3.5 and GPT-4 responses suggested using race in calculating kidney function, a claim she calls untrue because race is a social construct.
“they said that we should use race in calculating kidney function because there were differences in muscle mass between the races”
1:10:57 · Roxana Daneshjou · LLM bias
In a 2024 study, Roxana Daneshjou's team found 65 percent of surveyed dermatologists reported using large language models in their practices.
“65 % of dermatologists reported using large language models in their practices”
1:13:04 · Roxana Daneshjou · Clinician AI use · 2024
Stanford held a red teaming event, where participants deliberately stress-test AI, bringing together 80 physicians, computer scientists and engineers to probe language models with realistic medical tasks.
“we held this red teaming event at stanford where we brought together 80 physicians computer scientists engineers”
1:14:00 · Roxana Daneshjou · Red teaming
Roxana Daneshjou says ChatGPT flagged an incorrect medication dose in her daughter's after-visit summary after her daughter broke an arm.
“I gave it the whole after visit summary and I said, are there any mistakes here? And it clued in that the dose of the medication was wrong.”
1:19:21 · Roxana Daneshjou · Medication error
In a dermatology study with Matt Groh, AI assistance increased diagnostic accuracy for both groups, but the performance gap between skin tones also grew.
“there was actually situations where like the accuracy increased um in both groups but the gap actually also increased”
1:23:04 · Roxana Daneshjou · Human-AI bias
Peter Lee says Geoffrey Hinton predicted that by around 2018 human radiologists would no longer be needed because of AI, which did not come to pass.
“Jeff Hinton, Turing winner and one of the grandfathers of the AI revolution, famously had predicted many years ago that by 2018 or something like that, we wouldn't need human radiologists because of AI.”
1:24:48 · Peter Lee · Radiology prediction · 2018