Majority of Americans Would Not Rely on AI-Based Healthcare

A recent survey from Pew Research Center found that although there are certain areas where patients see the benefits of artificial intelligence (AI) use in healthcare, the majority are skeptical due to various factors such as potential inaccuracies and the impact on patient-provider relationships.

Through this survey, researchers from the Pew Research center intended to gather public views on AI in healthcare. They polled 11,004 US adults from Dec. 12 to 18, 2022.

A majority of patients (75 percent) noted concern about the speed at which AI is implemented into healthcare, especially when knowledge about how it may affect patients is limited. 

Thus, most respondents are not comfortable with AI use in healthcare. Around 60 percent of US adults expressed discomfort when asked for their opinion on a provider relying on AI for diagnoses and treatment recommendations.

Researchers also found that 57 percent of survey respondents believe using AI would worsen relationships between patients and providers, and only 13 percent believed it would improve it.

Further, health record security also stood out as a concern, with 37 percent stating that AI in healthcare would lead to poor data security. Conversely, 22 percent believe it would improve health record security.

Overall, more Americans (40 percent) believe that using AI would reduce provider mistakes, compared with 27 percent who think AI use could increase the number of mistakes.

But the survey reveals that respondents are split over the question of whether AI can positively impact care quality. Thirty-one percent think using AI would improve care for people like themselves, while 30 percent say it would make care worse, and 38 percent say it wouldn’t make much difference.

Also, patients with certain demographic characteristics expressed more trust and openness to using AI in their healthcare. For example, men, patients aged 18 to 29, and those with higher education levels like postgraduate degrees were all less skeptical about using AI, with 46 percent, 44 percent, and 47 percent, respectively, feeling comfortable with its use.

On the other hand, 66 percent of women, 64 percent of those over 65 years, and 63 percent of those with less than a high school degree say they are uncomfortable with AI use in their healthcare.

Among the subset of patients who experienced are concerned about biases related to race and ethnicity in healthcare, more are optimistic about AI's positive impact on the issue rather than pessimistic.

Among Black, Hispanic, English-speaking Asian, and White adults, 64 percent, 42 percent, 39 percent, and 27 percent, respectively, indicated that bias linked to race and ethnicity is a major problem in healthcare.

However, among the total population that noted these concerns, 51 percent believed that AI could help combat this issue. Fifty-four percent of White patients, 40 percent of Black patients, 50 percent of Hispanic patients, and 58 percent of Asian patients stated that relying on AI can help reduce biases.

Research has shown AI's ability to predict race, with one study published last May showing that AI can predict race accurately using medical imaging, while clinicians cannot. 

Among the AI models that accurately detect racial identity, researchers noted that the mechanisms they use to do so are unclear to researchers. This indicates that AI models already in use may accidentally exacerbate health-related disparities.