Survey finds Americans wary of AI for healthcare diagnoses and insurance claims

Support for limited AI tasks

A majority of Americans surveyed were comfortable with AI translating patient notes or scheduling appointments. Respondents accepted those tasks either with or without human oversight. More than half were also open to AI analyzing medical scans and lab results, as long as a healthcare professional supervised the process.

Resistance to decisions by AI

Respondents were less comfortable letting AI make consequential decisions. Almost half said they did not want an AI model to accept or reject an insurance claim, and 41% opposed AI making a diagnosis on its own.

Overall, 60% said they were uncomfortable with doctors using AI for diagnosis and treatment. Thirty percent were comfortable with it, while 10% were unsure.

Patients want clear disclosure

Galen Stocking, Pew Research Center’s associate director of science and society research, said people see value in using AI in healthcare but want to know when and how providers and administrative staff use it. Stocking told CBS News that people consider it extremely important for providers to disclose their use of AI. He also noted that 46% of respondents did not know whether AI had been used during their care.

Views on care quality are mixed

Americans were divided over whether AI will improve medical care. Thirty-seven percent said it would improve health outcomes, while a nearly equal share thought it could lower care quality. Stocking said concerns include the patient-doctor relationship and the security of personal health information.