ChatGPT, the OpenAI artificial intelligence capable of conversing with users, has demonstrated its great multitasking capabilities. It even became a problem for universities noticing that his students used this tool in final evaluations. Based on this, some researchers were able to test the AI to know its performance.
Great has been the discovery by the evaluators when they saw that the chatbot obtained surprising results. And it is that the tests that ChatGPT faced are those that are required in the United States to be a doctor or lawyer.
LOOK: Google asks its founders for help to improve the strategy to combat ChatGPT
Ethan Mollick, professor of economics at Wharton, has compiled the findings of the investigations in a tweet.
One of the exams he took was the USMLE (United States Medical Licensing Exam), where the AI managed to complete the three tests necessary to pass the exam. As indicated by the results, the chatbot approached the pass threshold on all three exams without the need for any preparation.
On the other hand, Christian Terwiesch, a researcher at Wharton University, explains that ChatGPT “has demonstrated the ability to perform at a professional level in tasks such as writing software and preparing legal documents” referring to the performance of artificial intelligence another test, that of the operations management course (MBA).
SEE ALSO: How to talk to ChatGPT, the popular artificial intelligence chatbot, from WhatsApp (in just four steps)
ChatGPT achieved a grade between a B or B- in this last evaluation. Although the AI erred on some questions, the researchers were surprised by the short time it took for the system to finish each exam.
In the case of the ‘Bar Exam’, the test taken by aspiring lawyers in the United States, the result was also remarkable, taking into account that a common person needs about 7 years of higher education and specific training to correctly answer an average of 68% of the questions. The OpenAI chatbot has achieved a score of 50.3%.