By Aijaz Alam
How many doctors and hospitals use AI?
72% of US physicians said they use at least one AI tool in their practice in January-February 2026, in the American Medical Association's survey of 1,692 physicians, up from 48% in 2024 and 38% in 2023. For hospitals, 71% of US non-federal acute care hospitals used predictive AI in their EHR in 2024 (ASTP).
- 72%US physicians using at least one AI tool in their practiceAMA survey, Jan-Feb 2026, up from 48% in 2024
- 81%Awareness or use of AI (AMA headline)AMA, 2026: adds 9% unsure which AI tools their practice offers
- 39%Physicians using AI for research summaries, the top useAMA, 2026, up from 13% in 2024
- 19%Physicians using AI to draft patient portal repliesAMA, 2026, up from 9% in 2024
- 71%US hospitals using predictive AI in their EHRASTP from AHA IT Supplement, 2024, up from 66%
- 59%Small hospitals (under 100 beds) using predictive AIASTP, 2024, against 96% of large hospitals
- 20.7%US health care and social assistance employer businesses using AICensus BTOS, Nov 2025-Feb 2026, against 17.9% of all US employer businesses
The numbers
The full breakdown
US physicians and AI in their practice, 2026
| Answer | Share of physicians |
|---|---|
| Use at least one AI tool in their practice | 72% |
| Unsure which AI tools their practice offers | 9% |
| No AI use in their practice | 19% |
| Awareness or use (AMA headline: use plus unsure) | 81% |
AMA question: which of 17 AI use cases do you currently incorporate into your practice? n = 1,342.
What US physicians use AI for, 2024 and 2026
| Use case | 2026 | 2024 |
|---|---|---|
| Summaries of medical research and standards of care | 39% | 13% |
| Discharge instructions, care plans or progress notes | 30% | 20% |
| Billing codes, medical charts or visit notes | 28% | 21% |
| Chart summaries | 28% | 12% |
| Draft replies to patient portal messages | 19% | 9% |
| Translation services | 18% | 14% |
| Assistive diagnosis | 17% | 12% |
Top seven of 17 use cases. n = 1,342 (2026), 1,183 (2024). AMA added two use cases in 2026, so waves are not strictly comparable.
US hospitals using predictive AI in their EHR, by hospital type, 2023 and 2024
| Hospitals | 2023 | 2024 |
|---|---|---|
| All non-federal acute care hospitals | 66% | 71% |
| Small (under 100 beds) | 53% | 59% |
| Medium (100 to 399 beds) | 75% | 80% |
| Large (400 beds or more) | 90% | 96% |
| Rural | 48% | 56% |
| Urban | 77% | 81% |
| Critical access hospitals | 46% | 50% |
| Independent hospitals | 31% | 37% |
| Members of a multi-hospital system | 81% | 86% |
Predictive AI built into or launched from the EHR (machine learning or risk models). n = 2,425 (2023), 2,080 (2024).
What hospitals use predictive AI for: the fastest-growing uses
| Use, among hospitals using predictive AI | 2023 | 2024 | Change |
|---|---|---|---|
| Simplify or automate billing | 36% | 61% | +25 points |
| Facilitate scheduling | 51% | 67% | +16 points |
| Identify high-risk outpatients for follow-up care | 78% | Not stated | +9 points |
n = 1,696 (2023), 1,587 (2024). For high-risk outpatients ASTP gives the 2023 share and a rise of 9 percentage points, not the 2024 share.
US employer businesses using AI, Nov 2025 to Feb 2026
| Businesses | Used AI in the last two weeks |
|---|---|
| Health care and social assistance (NAICS 62) | 20.7% |
| All US employer businesses | 17.9% |
Census BTOS AI supplement, pooled waves 17 Nov 2025 to 8 Feb 2026. A different period from the biweekly BTOS series.
Over time
US physicians using AI in their practice, every AMA survey
| Year | Use at least one AI tool | Awareness or use (AMA headline) |
|---|---|---|
| 2023 | 38% | Not measured (the 'unsure' option was added in 2024) |
| 2024 | 48% | 66% |
| 2026 | 72% | 81% |
AMA states 38% for 2023 in its 2026 report. It changed the question between waves: from 2024 it separates physicians who are unsure what their practice offers, and 2026 lists 17 use cases instead of 15. AMA says this limits strict comparability across waves. n = 1,081 (2023), 1,183 (2024), 1,342 (2026).
Why sources differ
Same question, different answers
| Number | When | Who measured | Why it differs |
|---|---|---|---|
| 72% | Jan-Feb 2026 | AMA physician survey | Named at least one AI use case in their practice |
| 81% | Jan-Feb 2026 | AMA physician survey (headline) | Adds 9% who were unsure which AI tools their practice offers |
| 38% | 2023 | AMA physician survey | Earlier wave without the 'unsure' option; AMA warns waves are not strictly comparable |
| 71% | 2024 | ASTP/ONC from AHA IT Supplement | Hospitals, not doctors: predictive AI built into the EHR |
| 20.7% | Nov 2025-Feb 2026 | Census BTOS | Health care and social assistance businesses that used AI in the last two weeks |
The figures range from about 21% to 81% because they count different things. The AMA's 81% is its headline, but it is 'awareness or use': it adds the 9% of physicians who were not sure which AI tools their practice offers to the 72% who named at least one use. The 72% is the cleaner answer to how many doctors use AI. Comparing it with 38% in 2023 overstates the jump a little, because AMA changed the question and added use cases between waves. The hospital figure of 71% is about institutions, not people: it counts hospitals with any predictive model built into the electronic health record, such as readmission risk or no-show prediction, and many of those tools run in the background without a doctor choosing them. The Census figure of 20.7% is lowest because it covers every health care and social assistance employer business, including small clinics, home care and child care, and asks only about the last two weeks. None of these counts patients who ask chatbots about their health.
What it means
How to use these numbers
AI has moved from a pilot to a routine part of practice for most US doctors in about three years. Explicit use went from 38% to 72%, and the average physician now uses 2.3 AI use cases, up from 1.1 in 2023. The growth is in paperwork and reading, not diagnosis: research summaries tripled to 39%, chart summaries more than doubled to 28%, and assistive diagnosis is at 17%.
Hospitals use a different kind of AI. Predictive models inside the EHR were in 71% of hospitals in 2024, and the fastest growth is administrative: billing (36% to 61%) and scheduling (51% to 67%). Among hospitals using predictive AI, 82% evaluated it for accuracy and 74% for bias in 2024.
The gap that matters is size. 96% of large hospitals use predictive AI, but only 59% of small ones, 56% of rural ones and 37% of independent ones. For a small practice or clinic, AI is likely to arrive through the EHR vendor and documentation tools, not through a separate project.
Method and sources
How it was measured
Method
- Publisher
- American Medical Association, 2026 Physician Survey on Augmented Intelligence (physicians); ASTP Data Brief No. 80 (hospitals); US Census Bureau BTOS AI supplement (businesses)
- Sample
- AMA: 1,692 US physicians, 15 Jan to 2 Feb 2026, n = 1,342 for the use question. ASTP: AHA IT Supplement, 2,080 non-federal acute care hospitals in 2024, response rate 51%.
- Definition
- AMA: physician currently incorporates at least one of 17 AI use cases into practice. ASTP: hospital uses machine learning or other predictive models embedded in or launched by its EHR.
- Released
- AMA: 12 March 2026. ASTP: September 2025. Census: 18 June 2026.
- Checked at source
- 31 October 2026, in the AMA PDF and the ASTP Data Brief 80 PDF
- Next release
- Not announced for the AMA survey; the AHA IT Supplement runs yearly
- Last changed
- 30 Oct 2026
- Data files
- CSV · JSON (stable links)
Sources
- 2026 Physician Survey on Augmented Intelligence, American Medical Association Industry body
- ASTP Data Briefs (patient portals; hospital predictive AI), HHS Assistant Secretary for Technology Policy / Office of the National Coordinator for Health IT Official data
- Business Trends and Outlook Survey, US Census Bureau Official data
Cite this page
APA
MLA
HTML link
FAQ
More questions people ask
Is it true that 81% of doctors use AI?
Not quite. 81% is AMA's 'awareness or use' figure for 2026. It includes 9% of physicians who were unsure which AI tools their practice offers. 72% named at least one AI use.
What do doctors use AI for most?
Summaries of medical research and standards of care (39% of physicians in 2026), then discharge instructions and care plans (30%), and billing codes or visit notes (28%).
Do small hospitals use AI?
Less than large ones. 59% of hospitals with under 100 beds used predictive AI in their EHR in 2024, against 96% of hospitals with 400 or more beds.
Every number on this page was checked at its original source (American Medical Association, HHS Assistant Secretary for Technology Policy / Office of the National Coordinator for Health IT, US Census Bureau). Free to cite with a link to this page.