Written By: Prerna Mittra | Updated : December 11, 2023 11:04 AM IST
Photo: Freepik
ChatGPT is all the rage right now, but can it accurately answer all medical questions? Launched on November 30, 2022, ChatGPT is a software -- or a chatbot -- developed by OpenAI. It is based on a 'large language model' that enables users to start and steer a conversation with the chatbot on absolutely any desired topic, in any length, style, format and language. But, a recent study has concluded that it may not be all that accurate when it comes to answering specific medical questions.
According to a CNN report, researchers at Long Island University posed 39 medication-related queries to the free version of the AI chatbot. The questions were real from the university's College of Pharmacy drug information service. ChatGPT's answers were later compared with responses "written and reviewed by trained pharmacists".
In the study, it was found that ChatGPT provided "accurate responses" to only about 10 of the questions posed. For the remaining 29 prompts, the answers were found to be "incomplete or inaccurate", or they did not address the questions, stated the report. The findings were presented last week at the annual meeting of the American Society for Health-Systems Pharmacists in Anaheim, California.
Sara Grossman, an associate professor of pharmacy practice at Long Island University and one of the study authors was quoted as telling CNN that because it is so popular, the researchers were concerned that their students, ordinary consumers and even pharmacists would turn to ChatGPT to "explore questions about their health and medication plans".
Per the report, the answers to the queries often yielded "inaccurate, or even dangerous, responses". Grossman explained that in one instance, when asked about taking two specific medicines together -- a Covid-19 antiviral medication and blood-pressure lowering medication -- ChatGPT responded that these two medications, when taken together, will not cause any "adverse effects" in the body, when in reality, patients people who take both medications often have a "large drop in blood pressure, which can cause dizziness and fainting". Clinicians, therefore, create "patient-specific plans", which includes lowering the dose of the blood-pressure lowering medication or "cautioning the person to get up slowly from a sitting position".
Earlier this year, ChatGPT cleared the US medical license exam. TheHealthsite had previously reported that researchers found that the smart chat-box learns from errors and upgrades itself. It means just like a medical student, the medical search on ChatGPT may benefit from more exposure and increasing user interaction. The other thing is that the chat box performs well with technical language which, for a medical practitioner, may sound useful, but for a layman, it may be difficult to keep up with the complex medical jargon, making them struggle to get the right answers.
Going back to the latest study, when the researchers asked the chatbot to provide scientific references to support its responses, the software could only purportedly do it for eight questions. Interestingly, it was also found that ChatGPT was "fabricating references". "The responses were phrased in a very professional and sophisticated manner, and it just seemed it can contribute to a sense of confidence in the accuracy of the tool. A user, a consumer, or others that may not be able to discern can be swayed by the appearance of authority," Grossman was quoted as telling CNN.
Disclaimer: The content on TheHealthSite.com is only for informational purposes. It is not at all professional medical advice. Always consult your doctor or a healthcare specialist for any questions regarding your health or a medical condition.
