While many are excited about the potential of ChatGPT, medical doctors are not jumping on the bandwagon, especially when it comes to using the AI (artificial intelligence) chatbot for health advice. According to a report, ChatGPT is unreliable and makes up fake information.
Doctors Are Against ChatGPT for Medical Advice
A study revealed that ChatGPT fabricated health statistics when asked for information on cancer, so doctors are advising against utilizing it for medical advice, Daily Mail reported.
The AI chatbot only got one correct answer out of 10 inquiries about breast cancer screening. Additionally, the correct responses were not as "complete" as those gleaned from a quick Google search.
According to researchers, the AI chatbot occasionally exploited phony scholarly publications to back up its assertions.
It comes with warnings that users should take caution when using the software since it has a propensity to "hallucinate," or make up stuff.
ChatGPT was requested to respond to 25 questions by researchers from the University of Maryland School of Medicine regarding recommendations for breast cancer screenings.
Each question was posed three times because the chatbot is known to change its responses. Three radiologists with mammography training then examined the outcomes.
Eighty-eight percent of the answers were suitable and straightforward to comprehend. However, they cautioned that some of the responses were inaccurate or fake.
As an illustration, one response was founded on outmoded knowledge. It suggested postponing a mammogram that was needed four to six weeks after receiving the Covid-19 immunization. The recommendation was amended a year ago, encouraging women not to wait.
A recent emergency halt to the "hazardous" "arms race" to introduce the newest AI was demanded by over a thousand academics, professionals, and tech sector executives.
They cautioned that the competition among tech companies to create ever-more potent digital minds has been "out of control" and poses deep hazards to society and humanity.
ChatGPT Is Inconsistent, Makes Up Fake Journal Articles
ChatGPT's answers to the queries about breast cancer risks and where to get a mammogram were inconsistent. The study discovered that the replies were dramatically different each time the same question was presented.
According to study co-author Dr. Paul Yi, consumers should be aware that these are new, unproven technologies and should still rely on their doctor, rather than ChatGPT, for advice. He added that they also observed ChatGPT making up fake journal articles or health consortiums to support its claims.
ChatGPT, according to lead author Dr. Hana Haver, relied solely on the American Cancer Society's guidelines and did not include alternatives from the Centers for Disease Control and Prevention or the US Preventive Services Task Force.
The phenomenon, which AI experts call "hallucination," occurs when a chatbot that has been educated on an answer cannot discover that answer and confidently responds with a response that it considers plausible.
It repeatedly gives the incorrect response after that without realizing it is a creation of its mind.
However, Dr. Yi claimed, in general, the results were encouraging, with ChatGPT accurately identifying queries on breast cancer symptoms, risk factors, and recommendations for mammography frequency, cost, and age.
For him, the percentage of correct answers was "quite astounding." Also, he was glad that it comes with the additional benefit of summarizing information in an easily digestible manner for customers to understand quickly.
The study was published on published in the journal Radiology.
Check out more news and information on Technology in Science Times.
