Do patients trust AI in healthcare enough to use it?

AI is helping patients become more informed and engaged participants in their healthcare journey. But as use grows, so do expectations around privacy, accountability, and human oversight.

AI confidence gaps based on patients’ gender, age, and location

The tendency of AI to hallucinate is well known, and 69% of patients say they are concerned by this possibility, with the number jumping to almost 80% for patients ages 25-29.

However, when it comes to accuracy, 81% of men say they are extremely/somewhat confident in the reliability of the AI-generated responses to their health questions, compared to 66% of women.

Similarly, younger patients between 18-34 are more confident that AI-generated responses to their health questions are accurate. This same younger patient population is also more inclined to share their personalized data with an AI health tool in order to receive a more customized response—provided their data is anonymized and not sold to a third party.

When looking at patients’ confidence in AI based on their location, 82% of urban patients say they are extremely/somewhat confident that the health answers they receive from AI tools are accurate, compared to 72% of suburban patients, and 65% of rural patients. Additionally, 32% of urban and 31% suburban patients say are willing to share their data to get personalized health advice if the data is anonymized and not sold to third parties, compared to only 19% of rural patients.

The great AI divide

While most patients feel moderately prepared to take advantage of AI healthcare tools over the course of the next three years, some clear distinctions emerge when it comes how they are willing to use it. Women are much more reluctant to engage with AI tools that require their personal information or health data than men. This holds true for rural patients, as well.

How prepared are you to take advantage of the following advancements in healthcare related to AI over the next three years?

Slide 5 - Bar Chart.svg
Slide 6 - Bar Chart.svg
However, regardless of location, patients express a strong interest in AI applications that help to simplify medical records and lab results into plain language, as well as AI tools that can help provide explanations of insurance benefits available to them.
Slide 7 - Pie Chart.svg

All health AI is not created equal

As AI in healthcare becomes increasingly more common, patients and clinicians – regardless of age, location, or gender – unanimously agree that it is important to have the sources and AI systems that are used to generate clinical content be validated by a human expert-in-the-loop.

Patients also expect that AI tools used by their doctor for clinical decisions are being explicitly approved by their health system to ensure safety and compliance with the rules (81%). In contrast to this expectation, a recent survey on Shadow AI that found that 40% of healthcare professionals say they have encountered unauthorized AI tools in the workplace, and nearly 20% admitted to using them.

Additionally, while patients understand AI can be an efficient way for their doctors to seek information related to their care, 78% of patients expect their doctors to confirm AI-generated information with other sources.

When it comes to patient-centered AI applications, such as AI health assistants, less than one-in-four patients say they trust AI assistants built by a major technology company or their local hospital/primary care provider. Interestingly, the exception is patients aged 18-24 who say they would trust AI assistants from a major tech company over AI built by their local healthcare provider.

Geographically, patients in rural settings say they are more apt to trust a specialized medical data base (27%) and their local hospital system (25%) for an AI assistant, distantly followed by major tech companies (17%), and trusting all AI equally (15%); whereas patients in urban settings say they trust all AI tools equally (28%), followed respectively by AI assistants from big tech companies (25%), local hospital AI tools (20%), and specialized medical databases (16%).

Importance of AI-generated health responses validated by human expert-in-the-loop (% very/somewhat important)

Slide 11 - Bar Chart.svg
Slide 12 - Bar Chart.svg
Back To Top