ChatGPT passes US MBA, Law, and Medical exams

Day after day, the ChatGPT has gained new uses and many people use the ability of the AI ​​that converses to create texts on the most diverse subjects. The novelty this time is that, according to a study carried out in the USA, the idea of OpenAI would easily pass complex tests that even students struggle to pass.

The highlight was the creation of legal documents with all the important details and the ChatGPT received a B grade when answering an MBA test. Understand the study right now.

ChatGPT passes MBA in USA

Ethan Mollick, professor at the University of Wharton in the United States, developed a study to find out how the AI ​​that talks OpenAI would do when put to the test in tests generally developed for professionals focused on managing people and projects.

When requesting content with an academic focus only, the teacher Christian Terwiesch mentioned that the artificial intelligence was approved with a “B or B-“ grade, which configures the approval. In the United States, grades vary from A to F (as if each letter had a weight of 2). The great highlight of ChatGPT it remained for the elaboration of legal documents with all possible details, knowledge that would make artificial intelligence a very competent professional.

The tool has demonstrated a remarkable ability to automate some of the skills of highly paid workers in general and, specifically, workers in positions held by MBA graduates, including analysts, managers and consultants.

Christian Terwiesch, professor at Wharton University

In tests involving people management, the tool was approved in the final discipline of the MBA Operations Management (operations management), giving correct and excellent answers. Even with some initial mistakes, the ChatGPT corrected itself and delivered remarkable solutions that are usually thought of by great leaders.

Interaction with humans was essential for the conversational AI to be able to deliver even more assertive answers, and the professor pointed out that in real time, when receiving tips from an expert on the subject, he did a “remarkably good” job.

In cases where it initially failed to match the problem to the correct solution method, the tool was able to correct itself after receiving an appropriate tip from a human expert. Considering this performance, Chat GPT3 would have received a grade from B to B- in the exam.

Christian Terwiesch, professor at Wharton University

AI has also passed tests in Medicine and Law

Another analysis to measure knowledge of the OpenAI was to put her through the same test as people who want to get approval to become Medicine and Law professionals. The highlight at this point was the excellent preparation of legal documents, including all the details that are usually forgotten by most candidates.

A ChatGPT managed to pass a test that is equivalent to the OAB Exam and confirms whether a person is ready to practice the profession of lawyer. The AI ​​score was verified by professors from the Universities of Michigan e Chicago.

When put to do the US Medical Licensing Exam, studies at the University of Yale proved that yes, the AI ​​has the knowledge to be approved and act like a doctor if it were a real person.

The tool developed two articles that would be approved in the USMLE (acronym for medical licensing exam), with a passing percentage of 50% to 60%. The average pass rate for candidates is 60%.

Ethics worries experts

It is clear that the advances are important and can help professionals in their daily lives, but the biggest fear of developers in the study is fraud. Since ChatGPT has been proven to pass special tests, this can be leveraged in scams, with students handing in content that was not originally written by them.

At the same time, professors at US universities claim that, at least in medicine, there is the possibility of using AI that talks for education and even in decision-making. But of course this should just be a support rather than the main factor.

The idea of OpenAI it still does not connect to the internet and all of its knowledge is based on facts and events that happened until the end of 2021. We can only wait to see how it will be used over the months of 2023. Would you believe that an AI would arrive so far away? Tell us us Comment!

See also other features

Learn how to use ChatGPT, the new internet sensation

With information: The Hill