ChatGPT | AI passes tests for doctors, lawyers and graduate students in the US

ChatGPT |  AI passes tests for doctors, lawyers and graduate students in the US

ChatGPT achieved the impressive feat of passing tests to practice the professions of doctors and lawyers. In addition, artificial intelligence technology managed to create a text that would receive a grade B — on a scale of A as the highest and F as the lowest — in an MBA program in the USA.

  • ChatGPT | How to identify texts produced by artificial intelligence?
  • Creator of ChatGPT and DALL-E plans AI that edits videos from text

The research was carried out by Professor Ethan Mollick at Wharton University. He found that the writing tool would be able to pass a basic MBA course where he teaches. The study was conducted for academic purposes and was intended to find out if the writing would be deep enough to match the ability of a human.

Mollick’s survey also mapped out other evidence to further delve into ChatGPT’s capabilities. According to the professor, the technology could pass the exams to practice the profession of doctor or lawyer. The software managed to produce legal documents, based on the good practices of North American Law, to the point of being able to satisfy the needs of the evaluators.

Approval in medicine and law

A Fortune magazine report states that the program has already successfully passed “The Bar” exam. In an applied simulation, the AI ​​was able to successfully reproduce the answer to the questions. The feat would have been proven by professors from the faculties of law in the states of Michigan and Chicago.

In medicine, studies by researchers at Yale University have proven that the program can pass the United States Medical Licensing Examination (USMLE). There are also at least two scientific articles that report the possibility of the technology being approved in the USMLE.

Going back to the Ethan Mollick study, ChatGPT performed with over 50% accuracy at all stages. In some situations, he increased the success rate to something close to 60% in the analyzed cases, a sufficient percentage to pass the exam.

As with the OAB in Brazil, the US also applies tests to measure the knowledge of aspiring doctors and lawyers. You have to get that license to practice there, so that means ChatGPT could be defending people in court or operating other bodies.

Fears of using ChatGPT

Taking tests instead of students is one of the biggest fears of teachers around the world in relation to ChatGPT — which can even offer a more “pumped” paid version. AI has already proven capable of writing summaries about scientific papers, academic articles, programming applications and several other complex activities.

In the case of tests for professional practice, the situation is even more delicate. Imagine if an unprepared student manages to “cheat” and pass a test like this with the help of artificial intelligence?

Currently, there is a great discussion about copyright involving the content produced by the robot. There are those who say it is necessary to give due credit, while others defend the total banning of the tool from the creative environment, which can be considered cheating by some people.

How will the academic environment deal with the threat posed by ChatGPT? Is there an action plan or is it too early to talk about a change in the way students are assessed? These are just a few unanswered questions that OpenAI’s technology leaves up in the air.