Indiana University School of Medicine's Melissa Kacena tested AI-powered chatbot Chatgpt to see if it can write medical reviews as accurately as humans. For this study, she gathered 9 students and the AI program into three groups and asked these groups to review and write an Article on a given topic. She told the first group to write their own review article and for another group, she told ChatGPT to write the article on the same topic. She told the third and last group to write an article with the help of ChatGPT. After that, she compared all the articles and asked her students and colleagues to fact check each information in the article.
The results suggested that the articles written by ChatGPT were easy to read and were overall simple as compared to the articles written by students. But 70% of the references in the articles written by ChatGPT were wrong and were mixed up from different studies. There were also hints of plagiarism in the articles written by ChatGPT.
Melissa Kacena said that the articles written by ChatGPT were so convincing and followed the proper syntax that it was hard to believe that the information was false. If the colleagues and students wouldn’t have fact checked everything, we could’ve believed that ChatGPT is right. ChatGPT also had better grammar than students but it used the tone that is not mostly seen in medical articles.
- Also read: Tech Analyst Dan Ives Draws Parallels Between Nvidia's Rise and the 90s Internet Explosion
Read next:
How To Get A Free US Number In 2024.
Unable To Find The Published Date Of A Web Page? Here Are 6 Hacks To See It.
