There’s a term for when you take medical advice from a search engine instead of a medical provider: Dr. Google. It often delivers worst-case-scenario diagnoses, spouts bad advice, and is the bane of real doctors who have to debate its merits with their patients. But there's a challenger to Dr. Google on the horizon, which might work with physicians instead of against them: Dr. AI.
AI chatbots have been getting all sorts of new jobs lately, and now they could assist physicians—although Americans are largely hesitant to trust healthcare AI, according to a Pew Research Center survey. The prospect of healthcare providers relying on artificial intelligence to diagnose diseases and recommend treatments was met with an unfavorable reaction by 60% of those surveyed, who said they would be uncomfortable with such a scenario.
Only 38% said it will lead to better health outcomes, 33% said it would lead to worse outcomes, and 27% said it wouldn’t make much difference. Here, the move-fast-and-break-things tech motto comes up against the first-do-no-harm ethos of medicine, with 75% of people expressing concern that healthcare providers will implement AI before fully understanding the risks.
One of the primary concerns about AI’s involvement in healthcare is the view by 57% that it would erode the doctor-patient relationship. This relationship is especially important when it comes to mental healthcare. Most of those surveyed (79%) said they would not want to use an AI chatbot if they were seeking mental health support. (While there are many options to obtaining therapy online, AI is not designed to provide such services.)
Additionally, people expressed distrust regarding privacy, with 37% worried that AI in a healthcare setting would jeopardize health-record security.
Some survey respondents said they look forward to positives, such as a potential reduction in medical errors and bias. Forty percent said they believed the incorporation of AI will reduce the number of mistakes made by healthcare providers, particularly for things like skin-cancer screenings.
Most of those surveyed (65%) said they would definitely or probably want AI to be used for their own skin-cancer screening. About half (55%) said they believed it would make skin cancer diagnoses more accurate, 13% said it would lead to less-accurate diagnoses, and 30% said it wouldn’t make much difference. When the results were examined more closely, though, Black adults (57%) were less likely than White (65%) and Hispanic (69%) adults to want AI involved in skin-cancer screening.
This brings us to the very real and concerning issue of racial bias in medicine. It was reported as a major problem in health and medicine by those who are Black (64%), Hispanic (42%), Asian English speakers (39%), and White (27%). And sure enough, the data sets used to train AI make it less likely to detect a skin cancer-based health threat for darker-skinned patients. So while 51% said that using AI would improve the problems of bias and unfair treatment, we are still confronted by the reality that AI is exactly as biased as the information we feed it.
The public’s distrust of AI when it comes to healthcare is understandable and has actually been borne out. A few years ago, IBM made much of the potential of its AI, Watson, in the field. Despite much fanfare, Watson’s foray into medicine was a failure, and it is now employed in customer service.