Letter: Rejecting Generative AI in Healthcare Will Harm Patients, Not Protect Them

In a strongly worded letter published in The Guardian, Dr. Robert Pearl argues that a failure to embrace generative AI in healthcare is not a cautious approach, but a dangerous one that will ultimately harm patients.

The letter serves as a direct response to a recent essay that raised concerns about the risks of generative AI in medicine. Dr. Pearl, the former CEO of Permanente Medical Group, one of the largest physician groups in the U.S., counters that while risks exist, they must be weighed against the preventable errors and deaths that plague the current healthcare system.

AI Fills the Gaps in Care
Dr. Pearl's central argument is that generative AI should not be seen as a replacement for clinicians, but as an essential tool to fill the dangerous gaps where clinicians are unavailable.

He argues that rejecting the technology means ignoring its potential to make medical care safer and more accessible. For example, generative AI can provide "timely, reliable guidance" to patients at home, while simultaneously alerting physicians to early warning signs or medication issues "before irreversible harm happens."

He also points to the significant benefits for mental health, where AI can offer support to individuals in the middle of the night when "anxiety or depression worsens and the only option is the emergency room."

"Both/And," Not "Either/Or"
The letter dismisses the "either/or" framing of "clinicians vs. AI." Instead, Dr. Pearl advocates for a "both/and" approach.

He concludes that the future of medicine is a combination of three elements: "dedicated clinicians, empowered patients and generative AI." He argues that this combination will lead to "safer, higher-quality and more accessible care than any of the three can achieve alone."

Dr. Pearl, who has previously stated that "AI will be as common in healthcare as the stethoscope," is urging policymakers and healthcare leaders to first acknowledge the failures of the current system before judging the risks of a technology that could help solve them.