Over 80% of physicians are using augmented intelligence to tackle documentation and other administrative tasks – more than double since 2023, with physicians feeling more positive about the technology than ever.
However, many still worry artificial intelligence (AI) could negatively impact patient safety – especially if patients use it to replace clinical judgment.
These findings come from the American Medical Association’s Center for Digital Health and AI which has conducted similar surveys since 2023. This year’s Physician Survey on AI found seven out of 10 physicians are optimistic AI will reduce clinician burden by streamlining administrative tasks. The most common applications of AI are for clinical care documentation and the summarization of medical research.
But 40% remain equally concerned about AI as they are excited, with respondents expressing fear the technology will negatively impact patient privacy and the patient-physician relationship. Nearly half of respondents worry patients may use AI for tasks requiring clinical judgement, such as the interpretation of radiology or pathology results.
These findings reflect recent Texas Medical Association research: TMA’s 2025 Health Information Technology (HIT) Survey found 56% of the state’s physicians believe AI will have both positive and negative impacts on health care.
“Yes, the future is incredibly bright – and yes, we need to continue to ensure our patients are treated in a safe and effective manner,” Philip Bernard, MD, told Texas Medicine Today. Dr. Bernard chairs TMA’s Committee on Health Information Technology and Augmented Intelligence. “I’m incredibly supportive of adopting this technology, but we need to be aware of their security concerns and the limitations.”
Dr. Bernard understands physicians’ hesitations, especially given AI’s propensity for “evolving so quickly.” He has a concern, too: that physicians working in hospitals must confirm that their own AI tools meet certain institutional guardrails before using them.
For example, “although these AI agents have incredible promise, they do not always have enterprise level security” that protects patients’ health information, Dr. Bernard said. Some platforms could share patient data with third-party developers, for instance, or not employ certain protocols that guard against data breaches.
In fact, 86% of AMA’s survey respondents say data privacy is an important key to facilitating widespread AI adoption.
Overall, however, physicians’ view of AI continues to trend positively. From 2024 to 2026, physicians’ uncertainty about AI fell from 18% to 9%, and 76% of physicians now believe AI tools give them an advantage in their ability to care for patients.
Per AMA’s survey, most physicians (85%) still want to be consulted or responsible for adoption of AI into their practice, and 92% want more education and training on AI. On that front, TMA is already ahead of the curve: the association’s annual conference, TexMed, will return this April with multiple CME courses on AI.
“I’m really impressed with the TMA team. We have some great opportunities coming up that make me incredibly excited,” Dr. Bernard said. “I just need to add one caveat: let’s use AI responsibly.”
Learn more about AI resources offered by the AMA by visiting the AMA STEPS Forward collection of digital health solutions. For more information about HIT, see TMA’s dedicated HIT webpage.