Openai released tool for recognizing its own texts

The tool was introduced a couple of months after Openai attracted the attention of society to Chatgpt – a bot that knows how to write texts as a person. Last week, Microsoft, which has already invested significant funds in Openai, announced multibillion -dollar investments to integrate the development of startup into its services.

Many educational institutions limited the use of ChatGPT from students, due to fears that the bot can harm the quality of learning. Sam Altman, the head of Openai, jokingly, said that it was also with calculators. However, he also stated that the company can help teachers recognize the “false” works done by AI.

The development of a new tool continues, but at present it still makes many mistakes. Openai is waiting for feedback from teachers and parents. According to Openai, the classifier program, when evaluating test texts in English, successfully identifies 26% of the texts created as “probably written AI”, but incorrectly classifies 9% of the texts created by a person written by AI

This is not the first attempt to create a tool for recognizing texts created by machines. Recently, a student of Princeton University Edward Tien presented GPTZERO for teachers. Openai previously released its tool for recognizing the texts created by AI in 2019, together with the LLM language model. According to Startup employees, the new version of the tool more effectively recognizes the texts that are issued for the texts written by man.

The tool is not effective in the analysis of texts shorter than 1000 characters, and Openai does not recommend using it for other languages ​​except English. Also, the created text can be easily corrected by hand, thereby complicating its identification. Openai already paid attention to 2019 that the identification of a synthetic text is a difficult task, but the company plans to continue to work on the recognition of the system in the future.

/Media reports cited above.