OpenAI develops tool to identify texts produced by artificial intelligence

A OpenAI, responsible for ChatGPT, launched this Tuesday (31) a tool to identify texts created by artificial intelligence. It would be the antidote to complaints from educators around the world who see the platform as a threat to learning.

titled “Ai text classifier”, the tool works as follows: you take suspicious text, which may have been created by AI, and paste it in the indicated space. From there, the site will check and classify the written material as: 

ADVERTISING

  • Very unlikely to have been generated by AI
  • Unlikely
  • Unclear
  • Possible or probable.

Understand:


Tool is in the testing phase

Despite the ease of consulting the text, the head of alignment OpenAI, Jan Leike, said in an interview with the Axios portal, in the United States, that there may be “false positives and false negatives”.

The person in charge of OpenAI It also says that, to reduce the chances of platform failure, it is recommended that the pasted texts have more than 1000 words, and that they are, preferably, in English. 

Concerns about the ChatGPT

Concerns surrounding content generated by artificial intelligence arose after educators and entities around the world exposed that the ChatGPT it was being used for activities that should be done by humans. New York schools, for example, even banned the use of AI.

ADVERTISING

A OpenAI he also says that he is developing other tools to identify materials produced by artificial intelligence. Including watermarks and improving the text checking site are alternatives.

OpenAI develops tool to identify texts produced by artificial intelligence (Reproduction OpenAI/Canva)

Learn more:

Scroll up