ChatGPT fails to give ‘satisfactory’ reply to nearly 75% of medication-related queries: study

Popular AI tool ChatGPT flubbed nearly 75% of questions about prescription drugs — with some responses causing potential harm to users, according to a new study.

Pharmacists at Long Island University posed 39 drug-related questions to OpenAI’s free chatbot — with only 10 of the responses deemed “satisfactory,” per the study, which was first reported by CNBC.

For the other 29 questions, the responses either did not directly address the question, were inaccurate or incomplete, according to LIU’s researchers, who presented the findings during the American Study of Health-System Pharmacists meeting in Anaheim, Calif., which began Dec. 3 and runs through Dec. 7.

“Healthcare professionals and patients should be cautious about using ChatGPT as an authoritative source for medication-related information,” said Sara Grossman, an associate professor of pharmacy practice at LIU and the leader of the study.

Grossman and her team pointed to a query about the relationship between the COVID-19 antiviral Paxlovid and the blood-pressure lowering medication verapamil as an example.