Half of consumers believe AI addresses health concerns as well as physicians

Nearly half of global consumers believe people proficient in artificial intelligence use can address health concerns using AI as well as or better than trained healthcare professionals. 

That’s a key finding from the 2026 Edelman Trust Barometer’s Special Analysis, produced in collaboration with the Yale School of Public Health. The school’s dean, Dr. Megan Ranney, said “it leads to lots of questions” about how, why, and how quickly to improve the accuracy of commercial large language models (LLMs). 

Consumer trust of AI to support health takes on added importance in the broader context of the survey. For example, 32% of consumers turn to content creators, spiritual leaders, or others lacking medical credentials help influence their health decisions – especially when it comes to diet, nutrition, and longevity. 

“The field of health information continues to get crowded,” said Yale researcher Dr. Matthew Facciani. “Doctors did not lose their perceived authority, but their voice got diluted in our noisy information ecosystem.”

Consumer perceptions of AI mirror medical literature

Hearing that 49% of consumers believe AI can outperform physicians may seem startling, especially when other surveys suggest patients have begrudgingly accepted AI’s role in healthcare. It’s worth noting that’s a net total of respondents who said “yes” to AI’s ability to perform at least one of the following four tasks:

• Triage care (26% of all respondents)
• Perform basic medical procedures (19%)
• Determine medication or treatment plans (19%)
• Diagnose illness or condition (16%)

As it turns out, consumer opinion broadly aligns with medical literature. For example, studies from the United Kingdom and the United States have demonstrated AI’s potential to improve efficiency in emergency department triage. Consumers turn to AI for “convenience, immediacy, attention, agency, access, and costs,” as healthcare economist Jane Sarasohn-Kahn wrote in her analysis of the Edelman-Yale report; clinical teams appear to as well.

On the other hand, researchers in Japan found AI models perform significantly worse than expert physicians in matters of medical diagnostics but are on par with physicians overall. As Sarasohn-Kahn put it, consumers still value physicians as the “source of truth” for diagnosis, treatment, and prevention. In the absence of experts, though – whether because of long wait times or a shortage of providers – consumers very well may turn to AI.

Reduce conflict between AI models and clinical guidance

The Edelman-Yale findings emerge as consumers and clinicians embrace AI tools but seek reassurance they can trust them to provide accurate outputs and keep data confidential. Provider organizations must also grapple with consumers trusting AI even though it’s wrong simply because answers sound authoritative.

That points to a broader trend in the Edelman-Yale report. Among global respondents who received conflicting recommendations from physicians and other sources, from AI models to non-credentialed influencers, just 51% always followed the physicians’ advice.

That reinforces Ranney’s point about LLM accuracy. If consumers increasingly turn to AI models, it’s imperative they align with what clinical evidence would conclude. This would reduce conflict between AI outputs and physician recommendations – and help AI stand out among the non-credentialed sources many consumers consult for medical advice. 

Ideally, clinical AI works best for prediction and narrow evaluation. (That explains its effectiveness for ED triage.) Meanwhile, AI models could overcome the common challenges of adapting to new situations through context-switching, or applying the information that’s most relevant to the current stage of treatment and adjusting outputs accordingly.

Above all, organizations can show consumers how to use AI. That means offering examples of focused questions that don’t require follow-up – something AI models still struggle with – and demonstrating how to verify outputs with information from trustworthy sources. Helping consumers see AI as a research assistant with readily identifiable limitations will better define AI’s role in clinical care and improve the consumer-physician relationship.