49% say AI users can match or beat physicians on some health tasks: Survey

 Nearly half of respondents asked about AI across 13 countries said people without formal medical training who know how to use AI could perform at least one health-related task as well as or better than a trained healthcare professional, according to a 2026 Edelman Trust Barometer analysis conducted in collaboration with the Yale School of Public Health.

The Edelman analysis was based on an online survey of 12,998 respondents across 13 countries conducted from Feb. 28 through March 11. The AI question was asked of half the sample. Results were weighted so each country contributed equally to the 13-market global average, and the margin of error for half-sample questions was plus or minus 1.6 percentage points at the 99% confidence level.

Here are six key findings from the survey:

1. The survey found 49% believed an AI-fluent consumer could perform at least one of four tasks as well as or better than a physician: determining whether someone needs to see a healthcare provider, performing basic medical procedures with access to the necessary medical equipment, determining the proper treatment or medication for a specific illness or condition, or diagnosing an illness or condition.
   
   
2. Twenty-six percent said AI users could perform triage as well as or better than a healthcare professional. Nineteen percent said the same about performing basic medical procedures, 19% about determining treatment or medication and 16% about diagnosing an illness or condition.
   
   
3. Views differed by age. Fifty-nine percent of respondents ages 18 to 34 said AI users could perform at least one of the four tasks as well as or better than a physician, compared with 52% of those ages 35 to 54 and 35% of respondents 55 and older. Among respondents with a university education or higher, 56% agreed, compared with 42% of those with less than a university education.
   
   
4. The findings come as healthcare professionals compete with other sources for influence over health decisions. Among respondents influenced by the sources studied, healthcare professionals were more likely than a weighted aggregate of uncredentialed sources to influence decisions about diagnosis, long-term treatment and disease prevention.
   
   
5. Uncredentialed sources were more likely to influence decisions about diet and nutrition and wellness and longevity. The aggregate included AI platforms, creators without medical training, friends and family, people with similar health conditions and spiritual leaders.
   
   
6. Among respondents who had received conflicting recommendations from a physician or healthcare provider and another source, 51% said they always followed the healthcare professional’s recommendation. Another 34% said their decision depended on the situation, while 15% said they always followed the other source. Thirty-seven percent of respondents met the criteria for that question.

https://www.beckershospitalreview.com/healthcare-information-technology/ai/49-say-ai-users-can-match-or-beat-physicians-on-some-health-tasks-survey/