When people worry that they\u2019re getting sick, they are increasingly turning to generative artificial intelligence like ChatGPT for a diagnosis. But how accurate are the answers that AI gives out? <\/p>\n
Research recently published in the journal iScience<\/a> puts ChatGPT and its large language models to the test, with a few surprising conclusions. <\/p>\n Ahmed Abdeen Hamed \u2014 a research fellow for the Thomas J. Watson College of Engineering and Applied Science<\/a>\u2019s School of Systems Science and Industrial Engineering<\/a> at Binghamton University \u2014 led the study, with collaborators from AGH University of Krakow, Poland; Howard University; and the University of Vermont.<\/p>\n As part of George J. Klir Professor of Systems Science Luis M. Rocha<\/a>\u2019s Complex Adaptive Systems and Computational Intelligence Lab<\/a>, Hamed developed a machine-learning algorithm last year that he calls xFakeSci<\/a>. It can detect up to 94% of bogus scientific papers \u2014 nearly twice as successfully as more common data-mining techniques. He sees this new research as the next step to verify the biomedical generative capabilities of large language models.<\/p>\n \u201cPeople talk to ChatGPT all the time these days, and they say: \u2018I have these symptoms. Do I have cancer? Do I have cardiac arrest? Should I be getting treatment?\u2019\u201d Hamed said. \u201cIt can be a very dangerous business, so we wanted to see what would happen if we asked these questions, what sort of answers we got and how these answers could be verified from the biomedical literature.\u201d<\/p>\n The researchers tested ChatGPT for disease terms and three types of associations: drug names, genetics and symptoms. The AI showed high accuracy in identifying disease terms (88-97%), drug names (90-91%) and genetic information (88-98%). Hamed admitted he thought it would be \u201cat most 25% accuracy.\u201d<\/p>\n \u201cThe exciting result was ChatGPT said cancer is a disease, hypertension is a disease, fever is a symptom, Remdesivir is a drug and BRCA is a gene related to breast cancer,\u201d he said. \u201cIncredible, absolutely incredible!\u201d<\/p>\n Symptom identification, however, scored lower (49-61%), and the reason may be how the large language models are trained. Doctors and researchers use biomedical ontologies to define and organize terms and relationships for consistent data representation and knowledge-sharing, but users enter more informal descriptions. <\/p>\n \u201cChatGPT uses more of a friendly and social language, because it\u2019s supposed to be communicating with average people. In medical literature, people use proper names,\u201d Hamed said. \u201cThe LLM is apparently trying to simplify the definition of these symptoms, because there is a lot of traffic asking such questions, so it started to minimize the formalities of medical language to appeal to those users.\u201d<\/p>\n