Eighty-one percent of physicians reported awareness or use of AI in practice in 2026, more than double the 38% who reported a single use case in 2023. Hospitals moved along the same curve, and 71% now run predictive AI inside the electronic health record. The FDA has authorized more than 1,600 AI-enabled medical devices for the U.S. market.
We traced every number below to the publisher that reported it, then fetched each source page to confirm the figure and the period it covers. Market size opens the report because it frames the money moving through the category. Physician adoption follows, then hospital adoption and the administrative work AI already runs inside a practice. The report ends with patient expectations for a provider using AI.
Key Takeaways
- Physician use of AI climbed from 38% in 2023 to 66% in 2024 and 81% in 2026, and the average physician now runs 2.3 AI use cases, up from 1.1.
- Hospital use of predictive AI reached 71% in 2024, though independent hospitals sit at 37% against 86% for hospitals inside a health system.
- Billing automation posted the fastest hospital gain of any use case, climbing 25 points to 61% in a single year, with scheduling up 16 points to 67%.
- 72% of U.S. adults call it extremely or very important that a provider disclose AI use, and 63% want more say in how a practice applies it.
How Big the AI in Healthcare Market Is in 2026
Money entering the category grows faster than almost any other health technology segment, and the device pipeline tracks that growth. Our analysis below measures how much money moves through it.
| Market Metric | Figure |
|---|---|
| Global AI in healthcare market, 2025 | $36.7 billion |
| Global AI in healthcare market, 2026 | $50.7 billion |
| Projected market, 2033 | $505.6 billion |
| Compound annual growth rate, 2026–2033 | 38.9% |
| North America share of global revenue, 2025 | 54% |
| Software solutions share of revenue, 2025 | 46% |
| Largest application, robot-assisted surgery, 2025 | 13% |
| AI-enabled medical devices authorized by the FDA, September 2026 | More than 1,600 |
Key Insights:
- North America generates 54% of global revenue, which makes U.S. reimbursement and disclosure policy the largest single variable in the forecast.
- No single application owns more than 13% of the market, so spending spreads across many narrow tools, with no dominant platform.
How Many Physicians Use AI
Physician adoption doubled in three years, and the share of physicians who could name no AI use at all fell to under one in five. Confidence moved with usage, since the number of physicians who see a patient care advantage climbed every wave. In our analysis below, the three survey waves measure the size of that shift.
| Physician Measure | 2023 | 2024 | 2026 |
|---|---|---|---|
| Report awareness or use of AI in practice | 38% | 66% | 81% |
| Average AI use cases per physician | 1.1 | Not published | 2.3 |
| See an AI advantage in caring for patients | 65% | 69% | 76% |
| Report no AI use | Not published | 34% | 19% |
| Unsure which AI tools are available | Not published | 18% | 9% |
Key Insights:
- The AMA surveyed 1,692 physicians in early 2026, and 81% reported awareness of or use of AI.
- 88% of physicians rate validation of safety and efficacy as important, and 86% say the same about data privacy assurances.
How Many Hospitals Run Predictive AI
Hospital adoption splits along ownership and size, and independent hospitals trail health system hospitals by 49 points. Smaller facilities gained ground in 2024 and still trail by the same margin. Our data below show how far adoption traveled by hospital type.
| Hospital Group | 2023 | 2024 |
|---|---|---|
| All non-federal acute care hospitals | 66% | 71% |
| Health system affiliated | 81% | 86% |
| Independent | 31% | 37% |
| Large, 400 or more beds | 90% | 96% |
| Small, under 100 beds | 53% | 59% |
| Rural | 48% | 56% |
| Critical access hospitals | 46% | 50% |
Key Insights:
- 96% of hospitals with 400 or more beds use predictive AI, against 37% of independent hospitals, a 59-point spread.
- 80% of hospitals get predictive models from the EHR developer, so the vendor contract decides what most practices can run.
Where AI Already Does Administrative Work
Front office and revenue cycle tasks posted the fastest growth of any hospital use case in 2024, and oversight grew with them. Scheduling and billing now reach more hospitals than treatment recommendations do. In our analysis below, the 2024 figures show what the work looks like today.
| Hospital Use or Oversight Measure, 2024 | Share of Hospitals |
|---|---|
| Facilitating scheduling | 67% |
| Simplifying or automating billing | 61% |
| Identifying high-risk outpatients, EHR-developed models | 88% |
| Recommending treatments, EHR-developed models | 53% |
| Evaluate models locally for accuracy | 82% |
| Evaluate models locally for bias | 74% |
| Monitor models after implementation | 79% |
| Run a committee or task force for predictive AI | 66% |
Key Insights:
- Billing automation climbed 25 points in one year, and scheduling climbed 16, the two fastest-moving use cases the survey tracks.
- 82% of hospitals check model accuracy locally, so four hospitals in five run a staffed review behind the software.
What Patients Expect From a Practice Using AI
Patients want a provider to say when AI enters the visit, and that demand climbs as the task moves closer to a diagnosis. Most adults also say the decision happens without them. Our analysis below sets the disclosure patients want against what they already know.
| Patient Measure, June 2026 | Share of U.S. Adults |
|---|---|
| Call disclosure of AI use extremely or very important | 72% |
| Want disclosure when AI makes a diagnosis | 81% |
| Want disclosure when AI analyzes a medical scan | 81% |
| Want disclosure when AI takes notes during a visit | 72% |
| Want disclosure when AI schedules an appointment | 56% |
| Say they have little or no say in how AI gets used | 53% |
| Want more say in how AI gets used | 63% |
| Have used an AI chatbot for a health reason | 34% |
| Know a provider has used AI on their care | 16% |
Key Insights:
- Demand for disclosure drops 25 points between a diagnosis and an appointment booking, so patients judge clinical AI and scheduling AI by different standards.
- Only 16% of adults know a provider used AI in their care, and 46% answered that they were not sure.
Further Reading
About This Report
This report was created for DocVA.
