WASHINGTON (dpa-AFX) - OpenAI has developed a tool to detect AI-generated text, but it is hesitant to use it due to concerns about potential negative user responses, according to a report by The Wall Street Journal.
'Our teams have developed a text watermarking method that we continue to consider as we research alternatives,' the tech company said in an updated blog post.
Notably, a survey by the AI company found that 59 per cent of middle- and high-school teachers believe students have used AI for schoolwork, up from 42 percent in the previous year.
The undisclosed tool, which has been ready for over a year, leaves watermark on the AI-written content, making it easier for another AI system to recognize it.
The tool could detect AI-generated text by ChatGPT with 99.9 percent accuracy, the report added.
However, an internal report revealed that nearly 30 percent ChatGPT users said that they would use AI platform less if the watermarking tool were deployed.
The ChatGPT-maker is also concerned about the possibility that the watermark could be easily removed with techniques like 'using translation systems, rewording with another generative model, or asking the model to insert a special character in between every word and then deleting that character - making it trivial to circumvention by bad actors'.
Moreover, it could 'disproportionately impact some groups,' the company warned. 'For example, it could stigmatize use of AI as a useful writing tool for non-native English speakers.'
Copyright(c) 2024 RTTNews.com. All Rights Reserved
Copyright RTT News/dpa-AFX
© 2024 AFX News
