ChatGPT, an artificial intelligence (AI) chatbot, has been the hot topic of discussion online for its conversational expertise ever since it made its debut in November. Users have been testing ChatGPT’s skills & power by asking it a variety of questions in a fun & light manner. The tool has also recently passed several notable exams, including the US Medical licencing exam, a Wharton Business School exam for the final exam of the MBA program’s operations management course, and four Constitutional Law exams from the University of Minnesota Law School.
Details about ChatGPT by OpenAI Passes the US Medical Licensing Exam
Today, Twitter CEO Elon Musk has criticised the AI chatbot after ChatGPT passed the US Medical Licensing Test. In response to a tweet concerning the capabilities of the chatbot, Elon Musk stated, “I’m sure everything will be great.”
Notably, the chatbot ChatGPT was developed by AI research startup Open AI, in which Elon Musk and Microsoft have invested money. Anyone who visits the website for the AI tool can use the chatbot to ask a question about any subject and receive a prompt, comprehensive response in paragraph form.
ChatGPT has proven its potential during the last few weeks. The tool has written quick and intricate articles, created poems and jokes, and even composed speeches for US congressmen. It has also created marketing proposals. There are worries that some human jobs could be replaced by AI, though.
But one of the AI tool’s most recent successes was passing the US Medical Licensing examinations. According to ABC News, the researchers looked at the chatbot’s theoretical top bounds in a pre-print study. They claimed that ChatGPT passed the US medical licensing exam (USMLE), one of the most challenging standardized tests available, with a score of above 50%.
Researchers put the program to the test by having it take a mock, condensed version of the USMLE, which all doctors must pass in order to be licensed to practice medicine in the US. The answers, which may have been open-ended written responses or multiple-choice questions, were fed into the AI tool by the researchers and were then independently graded by two medical adjudicators. Also, they made sure that the solutions to those queries weren’t already available in the dataset that the chatbot could access after training.
The researchers saw that ChatGPT performed at or close to the passing mark for exams without any specialized training or reinforcement, despite not having seen the answers beforehand. The tool scored higher than 50% on all tests and was getting close to the 60% pass mark for the USMLE. The report concluded, “Thus, ChatGPT is now well inside the passing range.”
