Researchers warn:

Beware of information from chatbots on medication

AI-supported search engines and chatbots do not provide reliable information on medication. The answers are repeatedly inaccurate, incomplete and often difficult to understand, according to a study, which therefore advises caution.

"A key finding of our study is that the quality of chatbot answers is not yet sufficient for safe use by users," explained lead author Wahram Andrikyan from the Institute of Experimental and Clinical Pharmacology and Toxicology at the University of Erlangen. "A clear indication that the information provided by the chatbot cannot replace professional advice is therefore essential in our opinion."

The starting point for the study was the experience that patients use the internet to find out about prescribed medication. In April 2023, the researchers therefore asked the AI-supported chatbot from the Microsoft search engine Bing ten common questions about the 50 most frequently prescribed medications in the USA, including questions about how to take them, side effects and contraindications.

"Risk for patients"
In general, the chatbot answered the questions with a high degree of completeness and accuracy, said Andrikyan. However, this was not the case for some questions. "This poses a risk for patients because, as medical laypersons, they cannot assess the accuracy and completeness of the AI-generated answers themselves," said Andrikyan.

The expert emphasized that there has been rapid progress in AI-supported search engines with an integrated chatbot function since last year's study. However, the improvements are not sufficient and the risks to patient safety remain for the time being. However, as chatbots are trained with different data sets, he believes it makes sense to investigate the safety and quality of other technical systems as well.