Americans have doubts about AI’s ability to improve health outcomes

Artificial intelligence's (AI) presence may be growing throughout the economy, but most

Americans are wary of the technology’s use in many aspects of health care, a new survey finds.

The survey, conducted by the Pew Research Center in December 2022, found that 60% of American adults would feel uncomfortable if their health care provider used AI for purposes such as diagnosing and recommending treatments for disease.

One of the main reasons for the discomfort is skepticism regarding AI’s effectiveness. Only 38% of those surveyed said using AI for diagnosis and treatment recommendations would lead to better health outcomes, while 33% felt it would lead to worse outcomes and 27% felt it wouldn’t make much difference. 

Opinions about AI’s impact on outcomes differed significantly by respondents’ gender, age and education level. Forty-five percent of men said AI would improve outcomes compared with 30% of women; 44% of those ages 18-29 said it would lead to improvements versus 34% of those 65 and older; and 50% of those with postgraduate degrees favored its use compared with 31% with no more than a high school education.

While there were also race- and ethnicity-based differences in views of AI, these were not as sharp as the other categories. Forty percent of Hispanic, 37% of white and 35% of Black respondents said AI would improve patient outcomes. Thirty percent of Hispanic and Black respondents, and 35% of whites, said AI would worsen outcomes.

Further apprehensions the survey revealed were over the speed of AI’s adoption and its impact on the doctor-patient relationship and the security of health records. Three in four respondents said they worried that providers would adopt AI technology too quickly and before fully understanding risks it might pose to patients. Conversely, 23% feared the industry will be too slow to use AI, thereby missing opportunities to improve patients’ health.

More than half (57%) of respondents said they feared AI’s use in diagnosing disease and recommending treatments would make the patient-provider relationship worse. Similarly, 37% felt using AI in health and medicine would make patients’ records less secure, compared with 22% who thought it would improve security.

The survey found two areas in which respondents were upbeat about AI’s use in health care: reducing medical errors and improving equity in outcomes. Forty percent thought AI would reduce the number of mistakes health care providers make, while 27% thought it would lead to more mistakes and 31% felt it wouldn’t make much difference.

And among the 70% of adults who identified racial or ethnic bias and unfair treatment as a problem in medicine, 51% think relying on AI for diagnoses and recommended treatment would lead to improvements, versus 15% who thought it would make the problem worse and 33% who felt it would stay about the same. 

In addition to general topics, the survey asked about views on four specific AI applications in use today or being developed for widespread use: AI-based tools for skin cancer screening; AI-driven robots that can perform parts of surgery; AI-based recommendations for pain management following surgery; and AI chatbots designed to support a person’s mental health.

Respondents were solidly opposed to AI’s use in three of the four areas, with the strongest opposition—79% in favor, 20% opposed—expressed to the use of AI chatbots to support mental health. The exception was in the use of AI in skin cancer screening, where 65% said they would “definitely” or “probably” want it and 55% said AI would make skin cancer diagnoses more accurate.