The technology of ChatGPT arrived to shake the market and leave us with our jaws dropping. This time, the overcoming went too far! The artificial intelligence of OpenAI, an American company, was summoned to carry out the test of Enem (National High School Exam) as well as thousands of Brazilian candidates take it annually.
ChatGPT carried out five-year Enem tests, from 2017 to 2021, with more than a thousand questions answered and obtained higher percentages than real candidates.
see more
Google develops AI tool to help journalists in…
Unopened original 2007 iPhone sells for nearly $200,000; know...
Learn about the yields of artificial intelligence in one of the most disputed races in the country.
The expectation was that the chatbot would reach 78.9% of the candidates in the test, and that's exactly what happened: the average achieved was 612.3 in objective tests.
Folha de São Paulo estimated that ChatGPT would outperform 98.9% of human candidates in human sciences and 95.3% in languages.
For these areas, for example, ChatGPT can only use existing data to be able to analyze the questions. Of course, this does not mean that artificial intelligence has the critical sense to solve the issues, as it only makes use of the previously addressed issues.
In the newsroom, the robot managed to reach a score of 700, higher than 68% of the candidates who scored 613 on the test average. The 2021 theme was used as a benchmark for the AI and was corrected by two education experts.
The errors were seen in the syntactic construction, in the addition of commas, it did not present data, it sinned in the sociocultural repertoire and it did not present the resolution for the presented problem.
In mathematical questions, the percentage reached surpassed only 27% of candidates to the Brazilian public university.
This smaller percentage was expected, as the exact questions were a real enigma for the OpenAI chatbot, as guided by the DeltaFolha data.
The question of mathematics arose from a very simple thought, being really one of the biggest questions regarding the use of technology.
As researcher Ricardo Primi stated, ChatGTP does not follow a line of reasoning and, therefore, cannot complete the mathematical question.
ChatGPT failures are levels that only humans can reach with logical reasoning and critical thinking. Compared to the candidates, the AI averaged the vast majority.
Lover of movies and series and everything that involves cinema. An active curious on the networks, always connected to information about the web.