Artificial intelligence, and worries that it will lead to patient misinformation, are key topics at this year's American Medical Association House of Delegates meeting in Chicago.
The six-day event starting Friday draws 3,000 physicians, residents and medical students. The group's committees and house of delegates will discuss and potentially vote on a wide range of policy proposals. Delegates will review three proposals tied to the excitement surrounding A.I.'s use in healthcare, including two focused on the technology's potential misinformation to patients.
Physicians and payers are investing millions of dollars to implement A.I. to streamline workflow, despite a lack of oversight and the technology's propensity for unintended outcomes.
At health systems, A.I. has become a potential solution for clinician shortages and a way to trim back-office costs. Rochester, Minnesota-based Mayo Clinic and Google Cloud announced a partnership this week to develop a generative A.I.-enabled enterprise search capability for providers and researchers. Executives at Boston Children’s Hospital are also bullish on A.I., including Open A.I.’s generative A.I. application ChatGPT, which Chief Innovation Officer Dr. John Brownstein likened to the iPhone or Google in terms of transformation.
In one proposal, delegate Dr. Albert Hsu, a reproductive endocrinologist at University of Missouri Healthcare, requests the association study how A.I. heightens the impact of public health misinformation, cyber-libel and slander and misrepresentation of physicians.
In another, the American Society for Surgery of the Hand and the American Association of Hand Surgery want to address how GPT tools can “fail at simple tasks” and “insidiously commit errors.” Those groups are asking the AMA to launch a study into the consequences of using GPTs and partner with federal regulators to protect patients from misleading A.I.-generated medical advice.
A.I. has been lauded as a more efficient way for payers to process claims. Multiple specialty care groups filed a joint proposal requesting the association push for more regulatory oversight on insurers' practices, including possible requirements for healthcare professionals to review claims before they are denied.
An AMA spokesperson said the association doesn’t comment on proposals before voting, which will take place Monday through Wednesday. Not every proposal is put up for a vote. The three A.I.-related proposals will likely be addressed as one discussion item, the spokesperson said.
In the past few years, the AMA has established three policies on augmented intelligence, different from artificial intelligence in that it still requires a human aspect. In those policies, the group has supported how augmented intelligence can advance patient care, but has also noted the need to set guardrails to avoid adverse patient outcomes and other liabilities and to incorporate these topics into medical education.
Other proposals at this year's meeting include removing body mass index as a standard measure of health, promoting race-conscious admission practices and setting guidelines to ensure equitable care access.