Physicians’ Attitudes Toward AI in Diagnosis and Treatment

By Campion Quinn, MD

Physicians’ views on artificial intelligence (AI) in healthcare are evolving from early skepticism to cautious optimism. Overall, surveys indicate that many doctors see clear benefits in using AI for diagnosis and treatment support, even as they voice concerns about trust, oversight, and integration. Below we summarize key findings from physician surveys, studies, and expert commentary on the acceptance, skepticism, and concerns surrounding AI in medical practice.

Growing Interest and Adoption of AI in Healthcare

Recent surveys show a trend toward greater physician acceptance of AI tools. The American Medical Association’s 2024 survey found 68% of physicians saw definite or some advantage to using AI, up from 65% the year prior [1]. Notably, two-thirds of physicians (66%) reported using some form of AI in their practice in 2024, a sharp rise from 38% in 2023 [2]. Physicians increasingly report that their enthusiasm outweighs their concerns about health AI: the portion whose excitement exceeded worries grew from 30% to 35% in one year [3]. (About 40% remain “equally excited and concerned,” reflecting a balance of hope and caution [4].)

Studies in different regions likewise report generally positive attitudes toward AI among doctors. For example, a multi-specialty survey in Bahrain found overall positive physician attitudes toward future AI applications, consistent with findings in other developed countries [5]. Another international survey noted that specialists like pathologists tend to be especially optimistic, viewing AI as an essential aid in their field [6]. In a Brazilian study, over 85–90% of physicians expressed favorable opinions on using AI to assist with interpreting exams, making diagnoses, and managing treatments, indicating strong support for AI’s assistive role in clinical tasks [7]. Younger physicians and those in tech-heavy specialties (e.g., radiology, pathology) often report higher familiarity and comfort with AI, though positive views are found across age groups [8].

Perceived Benefits in Diagnosis and Treatment

Physicians are particularly optimistic about AI’s potential to improve diagnostic accuracy and treatment planning. In a Kansas survey of frontline clinicians, over half (55.6%) agreed that AI-based tools can improve diagnostics, and a similar proportion said they would consider using AI in healthcare [9]. Likewise, a large survey of Korean doctors and trainees found 83% believe AI will be most useful in aiding disease diagnosis, and over half see value in AI for formulating treatment plans [10]. Doctors recognize that AI systems can rapidly analyze vast amounts of clinical data – a majority in one study cited AI’s ability to process high-quality data in real time as its greatest advantage [11]. Many physicians hope such tools can help catch findings that humans might miss, thus enhancing diagnostic accuracy and enabling more personalized treatments [12]. For instance, AI “digital eyes” in radiology and pathology have demonstrated higher detection rates for certain cancers when used alongside clinicians, bolstering physician confidence in diagnostic AI [13].

Physicians also appreciate AI’s potential to increase efficiency and reduce routine burdens. In surveys, the top cited opportunity for AI is often automating administrative tasks (like documentation or scheduling), which physicians say could free up more time for patient care [14]. “The AMA survey illustrates that physicians are increasingly intrigued by the assistive role of health AI to reduce administrative burdens, enhance diagnostic accuracy, and personalize treatments,” noted the AMA’s immediate past president [15]. Many frontline doctors share the hope that AI can assume data-heavy chores – from scanning records to drafting notes – thereby allowing clinicians to focus on clinical decision-making and patient interaction [16]. Renowned cardiologist Eric Topol echoes this optimism, predicting that AI will “free clinicians from the outsize burden” of clerical tasks and augment their ability to deliver precise, timely care [17].

Ongoing Skepticism and the Need for Human Oversight

Despite the optimism, physicians remain cautious and insist on retaining human oversight in AI-assisted care. Many doctors do not fully trust AI to operate autonomously in making diagnoses or treatment decisions without clinician involvement. In multiple surveys, physicians overwhelming agree that AI should serve as a support tool, with the physician supervising final decisions [18]. For example, a German survey noted 94% of people (patients and providers) favor AI as a doctor’s aid rather than a stand-alone diagnostician [19]. Correspondingly, in a Korean physician poll, 79% said that if an AI’s conclusion conflicted with their own, they would follow the human doctor’s opinion over the AI’s recommendation [20].

Sources

1. American Medical Association (2024). "Physician Survey on AI in Healthcare."

2. American Medical Association (2024). "AI in Medicine Report."

3. American Medical Association (2024). "Trends in Physician AI Adoption."

4. Banerjee et al. (2024). "Medical Doctors’ Perceptions of AI in Healthcare." Cureus.

5. Al Kaabi et al. (2023). "Physicians' Attitudes & Knowledge toward AI (Bahrain Survey)." BMC Med. Education.

6. Johnson et al. (2024). "Frontline Physicians’ Perceptions of AI (Kansas Survey)." JAMIA Open.

7. Oh et al. (2019). "Physician Confidence in AI (Korea Survey)." J. Med Internet Res.

8. Nunes et al. (2024). "Physicians' Expectations of AI in Daily Practice (Brazil Survey)." JMIRx Med.

9. Kawaguchi et al. (2022). "Acceptance of AI in Medicine (Japan Study)." JMIR Human Factors.

10. Eric Topol (2024). "NIH Lecture on AI in Medicine."