:quality(85)/cloudfront-us-east-1.images.arcpublishing.com/infobae/DYKDRSRH55ERPHPKLPAYPGQMUI.jpg)
Like every year in Peru, Medical students Those wishing to obtain a medical residency must undergo a series of examinations to obtain this professional degree, which in many cases proves to involve a degree of difficulty.
Before getting a title ResidentsDoctors must first complete a National Medical Examination (ENAM) Their qualifications make it easier for them to choose health centers to develop their capabilities Rural and Urban Marginal Health Service (Serums). However, not everyone is able to get a positive result, so one researcher turned to the technology to test its complexity.
So said Javier Flores Cohila, a professor and researcher in medical education at the university Southern Scientific University The lead author of the study, consulted him ChatGPT And other artificial intelligence (AI) machines if they can take the test, but they pretend to be medical students from Peru. The result will surprise you.
It should be noted that 43% of medical graduates from all Peruvian universities attended university Serums They refused I sleepAccording to the study, which analyzed observations from 2009 to 2019.
At Flores' request, it turns out that the A.I ChatGPT-4 (Latest version) Passed the National Examination of Medicine (ENAM) with 86% accuracy, in the 2022 exam.
“ChatGPT achieved a score of 17, and normally scoring more than 15 on ENAM is very unusual. Additionally, the AI’s performance increased if it was given specifications on how to approach the question. For example, when ChatGPT was asked to ‘pretend to be a doctor From Peru, “ChatGPT only answered 9 questions wrong,” Javier Flores Cohaila explained to Andina.
According to the researcher, the study relied on data from Inam 2022Which consisted of 180 multiple-choice questions. Thus, the AI result was compared with a sample of 1,025 medical students who took the same test.
The research team analyzed various factors such as question type, specific knowledge of Peru, discrimination, question difficulty and quality, and topic to determine their influence on incorrect answers.
The results revealed this ChatGPT It received 156 correct answers, while other similar AI applications succeeded with lower scores. For example, BingAI with 148; GPT-3, 120; cool, 120; and Claude, 118.
Javier Flores Cohaila He explained that this research does not seek to compare the IQ of medical students with artificial intelligence, nor does it seek to prove that artificial intelligence can replace the doctor; Because their work goes beyond just answering questions.
“Being a doctor is so much more than just giving Exam Licensing. The researcher stated that being a physician is the role we have of communicating with patients, collaborating with other health professionals, leading population health, promoting health, etc.
However, he stressed that these results could be encouraging Universities and research groups to assist in medical education.
As we remember, on December 3 Inam 2023. Since 2006, this exam has been a mandatory requirement for applying to the National Scheme of Medical Residents, which is why it provides a score.
In addition, this score also represents 70% of the score that is taken into account for the distribution of places in Rural and urban fringe health serviceknown as serums.
