company launches feature that identifies text made by robot

OpenAI, the company that created Chat GPT, launched a tool that seeks to identify whether a text was written by artificial intelligence or not. Chat GPT – an artificial intelligence system that generates texts in several languages, including Portuguese, suggests programming codes and solves mathematical problems – generated a debate about its impact on the teaching process.

This tool simulates, for example, a conversation with a human and generates specific answers even if the questions are more abstract. That is, the content can be customized and is easily accessible. The debate surrounding the system is that students can use the robot to do schoolwork, for example.

Due to these impacts on education, OpenAI launched a feature called “Classifier”, which indicates the probability that content, with at least a thousand characters, was created by a robot or a human. However, the tool itself pointed out some weaknesses in the system.

Classifier is not 100% accurate and can misclassify a text as having been done by artificial intelligence. Text generated by robots can also be easily edited to avoid Sorter. Furthermore, the tool may make errors if the text is written by children and is not in English, as the system was trained by content written by adults in English.