Artificial intelligence research startup OpenAI on Tuesday launched a tool that can determine if a text has been written by a human or generated by an artificial intelligence tool like ChatGPT, which is from the same company.
In a statement, OpenAI explains that, at the moment, it cannot guarantee 100% whether a text was created by a human or by a robot.
“In our assessments of a ‘challenge set’ of English texts, our classifier correctly identifies 26% of AI-written texts as ‘probably AI-written’, while incorrectly labeling 9% of AI-written texts by humans as if they were the product of artificial intelligence ”, details the company.
OpenAI noted that this tool will improve as the length of the text being parsed increases.
He also said that it only works on texts in English and that if a text generated by artificial intelligence has then been edited by a human, it can pose as a “human” text without being detected.
With this tool -they say- They seek to help the unions that could be most affected, such as educators.
“We recognize that identifying AI-written text has been a major point of discussion among educators and that it is important to recognize the limits and impacts of AI-generated text classifiers in the classroom,” the text notes.
Schools in the United States, for their part, have been quick to limit the use of ChatGPT for fear that the software could impair learning.
The launch of this new system comes two months after OpenAI captured the world’s attention by introducing ChatGPT, a chatbot that through artificial intelligence can coherently answer almost any question in a matter of seconds.
Tech giant Microsoft last week announced an investment “in the billions” in OpenAI and said it would incorporate the startup’s artificial intelligence models into its products.
MORE NEWS:
EFE International news agency based in Madrid and present in more than 110 countries.