Americans increasingly trust AI chatbots over doctors for health advice

The mechanics of AI prompts

The survey found that many users do not fully understand how AI generates its responses or how their own input affects the outcome. Nearly 78% of respondents failed to recognize that the specific way a question is worded can influence the answer the AI provides. This lack of awareness is significant because it can lead to biased or skewed information based on the user's phrasing.

Many users actively steer the AI toward a specific conclusion they are hoping to find. About 34% of respondents admitted to asking if a symptom could be connected to a condition they already suspected they had. Another 32% asked if a specific diagnosis could explain their experience, essentially seeking confirmation rather than an objective medical evaluation.

When users do not like the first answer they receive, more than half admit to reworded or adjusting their health question. This behavior shows that users are often seeking a specific type of reassurance or confirmation of a fear. Dr. Maggie Williams, medical director of virtual primary care at MD Live by Evernorth, noted that people often go to AI hoping for a certain answer.

Misconceptions about AI data

There is a notable gap between what AI actually does and what the public believes it can do. One in five respondents believed that AI only pulls information from trusted medical sources. Similarly, 22% of regular users believed the AI factors in their personal medical history even when they have never actually provided that information to the system.

Despite these misconceptions, many users are aware of their own limitations when interacting with the technology. Forty-two percent of respondents admitted they have even lied about who they were submitting a health question about for privacy reasons. Furthermore, 81% of users cited privacy concerns as the reason they leave out personal information when seeking guidance.

Because users often withhold their medical history and demographic details, the advice they receive is frequently incomplete. Nearly 59% of users later realized that the information they received from an AI did not fully apply to their specific situation. Despite this realization, nearly half of those users (49%) admitted they acted on that information anyway before discovering it was not the right fit.

Convenience and child health

The study highlights that AI users are highly active in their health management, with those using AI for health purposes asking an average of nine health-related questions every week. Users cite several reasons for this reliance, including the fact that AI is available at any time (37%), it is faster than seeing a doctor (31%), and it provides reassurance about health concerns (30%).

This trend is particularly evident among millennial parents who are using AI to manage their children's well-being. About 58% of these parents admit to acting on AI-generated health advice for their children without first confirming it with a healthcare provider. This group shows a high level of comfort in relying solely on AI for various parenting decisions.

The survey found that these parents are comfortable using AI to evaluate minor symptoms (80%), determine what type of care is needed (71%), and seek guidance on medication (71%). These figures suggest that parents are increasingly viewing AI as a primary tool for navigating the complexities of pediatric health and minor medical issues.

The need for human oversight

While the adoption of AI in healthcare is increasing, many users still harbor concerns about the risks of automated advice. About 34% of respondents expressed worry about receiving incorrect information, and 31% cited a lack of human oversight as a primary concern. These anxieties suggest that while the technology is useful, there is still a lingering fear regarding its reliability.

Looking forward, the study found that one in four people plan to use AI for more healthcare decisions in 2027. However, 27% of respondents stated they would trust AI-generated information more if it was reviewed by a doctor. This suggests a desire for a hybrid model where technology provides the initial information but a human professional provides the final verification.

Dr. Williams noted that people will continue to turn to AI for health questions and that having people more engaged in their health is a good thing. However, she emphasized that healthcare providers offer essential clinical guidance by catching details a patient might not consider relevant. Doctors are able to ask follow-up questions and put information into the context of a person's specific health history to ensure appropriate safeguards.