ChatGPT owner launches tool: OpenAI, the firm behind ChatGPT, made a software tool to detect AI-generated language available, according to a Wednesday.
Since its release in November, ChatGPT, a free tool that generates writing in response to a prompt, has been quite popular, raising questions about copyright and plagiarism. This material can include articles, essays, jokes, and even poetry.
The AI classifier, a language model trained on a dataset of pairs of texts on the same subject produced by humans and by AI, tries to identify texts created by AI. The company said that it employs a range of suppliers to solve problems like automated disinformation campaigns and academic dishonesty.
OpenAI warns that the classifier in its open beta version struggles with texts shorter than 1,000 characters and can be deceived by altered AI-generated text.
In order to collect input on whether imperfect tools like this one are helpful, OpenAI claimed that it was making the classifier available to the general public.
“We acknowledge that addressing the limitations and implications of AI generated text classifiers in the classroom is equally important as recognising that detecting AI-written text has been a significant point of discussion among educators.”
Some of the biggest US school districts, including New York City, have banned ChatGPT since it launched in November and amassed widespread popularity among millions of users due to worries that kids will use the text generator to cheat or plagiarise.
Schools use third-party detection programs like GPTZeroX to spot AI-generated writing.
In addition to working on the detection of AI-generated language, OpenAI said it is talking with educators about ChatGPT’s strengths and weaknesses.