Physicians are warming to AI, but want a seat at the table

Physicians are slowly but surely warming up to integrating artificial intelligence (AI) into their daily practices, but are still actively requesting more involvement and oversight before they’re ready to put their full trust in AI-enabled tools, according to a new survey by the American Medical Association (AMA).

In November of 2024, the AMA repeated its benchmark 2023 study to gauge changing attitudes toward AI across the physician community, and found that physicians are somewhat more likely to view AI favorably – and significantly more likely to be using it regardless of their personal feelings – than just a few years ago.

Key findings from the poll of more than 1100 physicians include:

• More physicians are letting their enthusiasm outweigh their concerns. In 2024, 36% of physicians reported feeling more excited than concerned about AI compared to 30% in 2023. The number of physicians expressing ambivalence (equally excited and concerned) hasn’t changed much, with 41% in 2023 vs. 40% in 2024.
• A growing number of participants recognize the potential benefits of AI. In 2024, 68% reported at least some advantage in patient care, which is a bump up from 63% in 2023. However, only 28% said there was a “definite advantage” to AI tools, with the remaining 40% believing there is “some advantage” to be gained with AI.
• Physicians are becoming more familiar with the use cases for AI in both the administrative and clinical environments, particularly around the use of generative AI to create documentation and assist with coding/billing, generate patient-facing communications, and produce chart summaries.
• As a result, they are more likely to believe that AI can improve work efficiency (75% in 2024 vs. 69% in 2023), reduce cognitive burdens (48% in 2024 vs. 40% in 2023), and relieve stress and burnout (54% in 2024 vs. 44% in 2023).
• Real-world usage is also on the rise. In fact, it has nearly doubled from 38% in 2023 to 68% in 2024. Top use cases remain largely in the administrative realm, and include documentation of billing codes, charts, and notes; creation of discharge instructions; summaries of medical research and standards of care; and translation services.
• Those that are not yet using AI are generally planning to do so very soon. By the end of 2025, close to 60% of respondents said they’ll be using AI for translation services and documentation/summary tasks. They’re also making moves into patient-facing areas, such as generation of draft responses in patient portals, patient-facing chatbots for customer service functions, and patient-facing health recommendations, and even assistance with diagnoses.

All of these factors seem to be adding up to a smooth transition into an AI-enabled world within just a few years. However, it might not be as simple as it sounds.

For one thing, physicians themselves aren’t in the driver’s seat when it comes to adoption decisions. The survey revealed that 65% of physicians say they have little to no direct influence when it comes to AI adoption decisions, and only 17% have “all or most” of the power in the room when decisions are being made.

Even more concerning, these numbers represent a drop from 2023 levels, where 20% of physicians were leading the decision-making process and 28% shared influence equally with others.

Pushing physicians from their seats at the table could be problematic, not least because it represents a significant disconnect from what physicians want. More than half (52%) expressed that they wanted to at least be consulted about AI adoption decisions, and 31% wanted to take on responsibility for selection and implementation of new tools.

Without involving physicians in the process more thoroughly, the industry may not be able to move the needle on a key criterion for successful adoption: user trust and acceptance of AI tools.

Right now, physicians may be showing enthusiasm for concepts, but they’re not entirely sure the AI ecosystem is set up in the right way just yet. When asked what would be necessary to increase trust in AI and the likelihood of adopting AI tools, more than 60% of physicians said they wanted increased oversight of AI-enabled software and devices by government agencies, including the FDA.

On a more local scale, participants remained steady in their desire for their organizations to implement meaningful policies, pathways, and guardrails for AI adoption, such as ensuring data privacy, liability coverage, safety, and clinical accuracy. Respondents also advocated for user-friendly integration with EMRs and other digital workflows, as well as clear and trustworthy evidence of safety and efficacy.

Keeping clinicians involved in the decision-making process will be an important strategy for ensuring that AI can be safely and successfully adopted, especially as AI-enabled tools start to move into the patient-facing environment.

With strong interest from physicians to be actively involved, organizational leaders may wish to explore how to bring clinical champions – not just physicians, but nurses and other practitioners as well – into the right conversations. Doing so may help to keep the positive momentum building and help to design and implement an AI-driven environment that produces value for all involved.

Jennifer Bresnick is a journalist and freelance content creator with a decade of experience in the health IT industry.  Her work has focused on leveraging innovative technology tools to create value, improve health equity, and achieve the promises of the learning health system.  She can be reached at [email protected].