Many people these days consider artificial intelligence as a threat to their jobs. Now it seems that maybe one day a robot doctor will be launched on the market. In a now peer-reviewed study, researchers have confirmed that the ChatGPT artificial intelligence was able to pass an important medical exam in the United States; However, this acceptance was not obtained with a high score.
A group of researchers asked ChatGPT to participate in the United States Medical Licensing Exam (USMLE) and this artificial intelligence scored between 52.4 to 75 percent gained. The USMLE exam is a three-stage standardized exam that doctors must pass in order to obtain a medical license.
Researchers say that the average passing score for passing this test is 60 percent, and ChatGPT to the first The artificial intelligence model that has been able to succeed in this difficult test. It should be mentioned that the researchers did not contribute anything to the artificial intelligence of OpenAI company in conducting this test. The results of this research have been published in the scientific journal PLoS Digital Health after peer review.
The responses from ChatGPT were overwhelming
Regardless of the average ChatGPT score, researchers have defined the capabilities of this artificial intelligence tool in producing correct and uncopied answers. Researchers say that this chatbot in 88.9 percent From his answers, he has provided new data that is medically correct. This tool also demonstrated the capabilities of deductive reasoning, chain of thought and reliability skills.
These findings appear to be unique to ChatGPT and its particular learning style. The OpenAI model, unlike other models that used deep learning, with the help of Large language models Exercise is given to predict a sequence of words based on the context of the previous words. In other words, ChatGPT can create sentences that have not been seen by the algorithm before but still make sense.
Previously, in an article, we reviewed some of the tests in which ChatGPT appeared successful. You can read that article through this link.