Recent news articles highlight the dysfunction of the American health care industry. Doctors are publicly feuding with insurers over denied claims; nurses are striking en masse for better staffing ratios and higher wages. And in the background, dissatisfaction among the public has resulted in widespread frustration and even violence. Physicians want more time and control over their practice. Could artificial intelligence be part of the solution?

This report analyzes responses to a survey about AI and health care fielded among a verified national sample of 839 practicing U.S. physicians. The survey was administered between July 16 and August 12, 2025, to assess how doctors are using artificial intelligence (AI), how they view the influence of health‑care stakeholders (especially insurance companies), and how they think system‑level factors — such as the lack of universal or single‑payer coverage — shape U.S. health outcomes. The margin of sampling error for population estimates is 3.4%. Uncertainty for demographic subgroups is larger.

1. Insurance companies are perceived to wield outsized influence

The survey results reveal near-universal concern among physicians about insurance company influence. When asked about the level of influence health insurance companies have on the U.S. healthcare system, 93% of respondents said insurers have "too much" influence. Only 5% believed the current level of influence is "about right," while a mere 3% said insurers have "too little" influence.

Similarly, 77% of physicians said that pharmaceutical companies have too much sway, while 19% said they had “about the right amount of influence.”

The finding underscores how deeply patients’ insurance policies shape both the care they get and the daily reality of clinical practice. Physicians across the board report that insurer rules constrain their clinical decision-making and that doctors themselves have too little influence in the healthcare system. Physicians working in emergency medicine and OB-GYN were the most likely to feel they had little influence, with 94% agreeing.

Figure 1

‍

These results suggest physicians have overwhelming concerns about whether the healthcare system is structured to prioritize patient outcomes and practitioners' advice over financial and other considerations.

2. Doctors see behaviors, costs as the biggest barriers to better outcomes

Physicians also connected systemic policy and institutional design choices to health outcomes. A strong majority (62%) believes that the lack of universal or single-payer health coverage contributes at least somewhat to the United States achieving worse health outcomes than peer nations. Within this group, 31% said the coverage gap contributes "a great deal" to inferior results. Physicians also identified unequal access to healthcare as a major driver of poor outcomes, with 44% saying it contributes "a great deal" and another 34% saying it contributes "somewhat."

The most significant influence, however, was that of general health behaviors, insurance company policies, and prices of care. 77%, 68%, and 69%, respectively, of physicians in our survey said these factors  “contribute to worse health outcomes in the United States relative to other advanced countries.”

Figure 2

‍

Taken together, these results suggest that physicians see structural reforms — particularly those that expand coverage/reduce costs, and that emphasize healthy lifestyles — as essential levers for improving population health. The medical profession's support for such changes represents an important voice in ongoing policy debates about the future of American healthcare; the differential focus on improving lifestyle health and extracting better incentives from insurers suggests mainstream debates about federal program design may be missing key variables that determine outcomes.

3. Nearly a majority of doctors report using AI today

Artificial intelligence has moved from a theoretical possibility to a practical reality in medical practice. The survey found that 46% of physicians currently use some form of AI in their work. Among current uses, administrative tasks lead (24%), followed by clinical decision support (21%), research (12%), imaging interpretation (10%), and patient engagement (7%). Only 2% say they are not sure if AI is used in their practice. 

Our survey suggests AI adoption is approaching a tipping point in health care — no longer confined to early adopters or academic medical centers, and on the road to universal adoption. The pace of adoption has been driven primarily by practical applications that address immediate diagnostic and clerical challenges.

Like in other industries, among physicians who have adopted AI, the applications are notably pragmatic. Administrative task lead uptake, with 24% of AI users reporting this as a primary application. These include documentation, scheduling, and billing, which can consume significant amounts of time that physicians can otherwise spend administering care.

Clinical decision support ranks second among adopters at 21%, suggesting meaningful uptake of tools that help with diagnosis, treatment planning, and risk prediction at the point of care. Research applications account for 12% of use cases, while imaging interpretation (10%) and patient engagement tools (7%) round out the top applications.

Figure 3

‍

This pattern reveals a profession focused on AI's near-term value proposition: reducing administrative burden and supporting clinical judgment with evidence-based recommendations. Physicians are not waiting for revolutionary breakthroughs; they are adopting tools that make daily practice more manageable and efficient today.

4. AI holdouts see tools as unnecessary, unreliable

Understanding the 54% of physicians who have not yet adopted AI is crucial for predicting future uptake and designing solutions that address doctors’ needs. The survey asked non-users (approximately 471 physicians) about barriers to adoption, revealing a diverse set of concerns.

The most common explanation — cited by 33% of non-users — is simply that AI is "not necessary for my practice." This suggests that many physicians do not perceive current AI tools as addressing problems they face, or that their specialty or practice setting does not lend itself to available applications.

Nearly as many non-users (32%) cite unfamiliarity with available tools, indicating a knowledge gap that education and outreach could address. This represents perhaps the most actionable barrier for AI companies, as it reflects insufficient awareness rather than active resistance.

Concerns about accuracy and reliability were cited by 19% of non-users, reflecting appropriate clinical caution about trusting algorithmic recommendations with patient care. Cost barriers (13%) and privacy/ethics concerns (11%) round out the major impediments.

Figure 4

‍

These findings suggest that expanding AI adoption will require a multi-faceted approach: demonstrating clear value for diverse practice types, improving physician education about available tools, addressing accuracy concerns through validation and transparency, clarifying organizational policies, and ensuring cost-effective implementation.

5. Physicians in their own words

5.1 What Physicians Hope AI Will Accomplish

When we asked doctors to tell us in their own words what they want from AI, three clear themes came up. The biggest is getting rid of paperwork and other administrative burdens: One respondent told us they would love an AI  system for doing paperwork in general. They want a tool that can “automate the administrative burden that has become synonymous with modern medical documentation and make health records systems communicate better.”

Hand in hand with that, doctors say they are drowning in documentation, particularly in getting authorization from insurance companies for treatment. One respondent said they’d like an AI to help “with tedious tasks such as documentation so physicians can spend more time helping patients. Prior authorization help would be awesome!”

Other hopes surround diagnostics and risk assessments — using AI to scan images of body and brain scans for abnormalities is an increasingly common use case. Some physicians also want help keeping up with research (because who has time to read every new study?). And others think AI could be great for patient education — giving people personalized health info and helping them stick to lifestyle changes.

Figure 5

‍

5.2 The Barriers Physicians Perceive

When we asked about barriers, doctors had plenty to say. Money is a huge issue, especially for smaller practices. One doctor put it perfectly: "until [clinicians, tech, and admins] are on the same page, adoption is going to be limited." It’s expensive to buy software, integrate it into existing systems, train staff on usage, etc.

Clinicians are also worried about liability. As one physician said, "Who is liable when AI makes a mistake is still undefined." And they are skeptical about whether AI is actually accurate (“The quality of AI is dependent on the accuracy and completeness of the data, and current systems are too prone to error and misinterpretation.”). A doctor can validate their assumptions and explain their reasoning, instead of just spitting out mysterious "black box" answers.

Lastly, privacy concerns are a major factor, too. Are AI systems HIPAA compliant? How prone are they to data leakage? Is it ethical to train algorithms on patient data — which is necessary to improve the accuracy of these systems? AI tools come with a non-zero amount of legal risk for doctors, a drawback that must be weighed against any improvements in efficiency or care.

Conclusion

This national survey captures American physicians at a point of massive potential technological and policy change. Doctors are managing the daily realities of insurance-dominated healthcare while beginning to integrate AI into practice and envisioning technology's potential to reshape medicine. Doctors are adopting AI where it provides clear value today while hoping it will eventually address deeper challenges around burnout, access, and care quality.

At the same time, the near-consensus that insurers have too much power and that coverage gaps harm outcomes suggests a demand for policy, not technological, change. The challenges the American healthcare industry faces will not be solved completely by automation or the offloading of physicians’ duties.

As healthcare continues its technological transformation, these physician perspectives offer essential guidance. Improvements in outcomes must address clinicians’ needs. The only way to understand those needs is to ask doctors directly.