ChatGPT he is good in humanities but slips in exact sciences; understand the test carried out with the Enem exam

The analysis considered the score of each course and showed that artificial intelligence would perform well in the Enem, but its performance in mathematics was considered low, obtaining an average of 443,1 points, below the 527,1 average of human candidates. On the other hand, the human sciences, curiously, were taken by technology. In the simulation, the AI ​​average was 725,3, higher than the 523,3 points of real competitors.

Methodology used Enem tests from the last five years

The assessment of the ChatGPT was based on AI responses from tests taken over the last five years, answering 1.290 questions. The methodology used in the test was Item Response Theory. This mathematical model adopted by Enem predicts items calibrated according to parameters of discrimination, difficulty and probability of a random hit, as narrated by DeltaFolha. 

A Sheet calculated the final grade of the ChatGPT, using Inep's standard analysis, in which the machine answered each question only once, without previous examples, indicating the alternative it considered correct.

Although he excelled in human sciences and languages ​​and codes, the ChatGPT presented low performance in mathematics, which could be an obstacle to entry into popular courses at the country's main federal universities. Even so, AI obtained an average score of 608,7 on the Enem, better than that obtained by 79% of students that year, when added to the essay grade.

Read also

Bill Gates questionthe letter that asks for a pause in the development of artificial intelligence

The founder of Microsoft, Bill Gates, took a stand against calls to pause the development of artificial intelligence (AI). In an interview with Reuters, Gates argued that stopping the tools would not solve the challenges presented by the technology. Instead, the philanthropist suggested finding better ways to use AI. The testimony comes after a group of experts, including Elon Musk, call for a six-month pause on the development of AI systems that exceed the capabilities of the company's GPT-4 OpenAI.

ChatGPT is blocked in Italy due to non-compliance with personal data legislation

The Italian authorities announced this Friday (31) the temporary blockade of the ChatGPT, accused of not respecting legislation on personal data and of not having a verification system for underage users. A OpenAI has 20 days to adopt measures to protect minors under penalty of a fine and increased lockdown. O ChatGPT arrived in Italy in November 2022 and was quickly adopted by users impressed by its ability to clearly answer difficult questions, write sonnets and even pass tests and exams.