(The Hill) — OpenAI, the company that launched ChatGPT, announced on Tuesday it has created a tool that can tell the difference between text generated by artificial intelligence (AI) and text written by a human.
The “classifier” the company created, by its own admission, is not always 100 percent accurate in distinguishing between the text.
“While it is impossible to reliably detect all AI-written text, we believe good classifiers can inform mitigations for false claims that AI-generated text was written by a human: for example, running automated misinformation campaigns, using AI tools for academic dishonesty, and positioning an AI chatbot as a human,” the company said in a blog post.
The announcement comes after ChatGPT rose quickly in popularity, especially among students, for its ability to give human-like responses to questions and turning the responses such as essays or emails.
The program also brought a wave of controversy in education as ChatGPT gives different answers to the same questions, making it impossible to tell if a student used ChatGPT to write their essay or assignment.
As a result, New York City and Seattle Public Schools banned the chatbot from their servers, citing concerns of cheating and a lack of critical thinking skills needed with the program.
The company responded by saying they were working with schools to ease their concerns.
The classifier made in response is “very unreliable” for catching AI-generated text for writing under 1,000 characters, but gets more reliable the longer the text is, according to the company.
“In our evaluations on a ‘challenge set’ of English texts, our classifier correctly identifies 26% of AI-written text (true positives) as “likely AI-written,” while incorrectly labeling human-written text as AI-written 9% of the time (false positives),” the company said.
The company added that it is launching the imperfect tool to get feedback and will continue to work on ways to detect AI-text.
It is unclear if tools like this will ease the concerns of teachers, with some students saying professors have already changed how assignments are conducted due to ChatGPT.
In the blog post, ChatGPT included a resource for educators to go over exactly what ChatGPT is and the types of uses and limitations it can have.
“We are engaging with educators in the US to learn what they are seeing in their classrooms and to discuss ChatGPT’s capabilities and limitations, and we will continue to broaden our outreach as we learn,” the company wrote. “These are important conversations to have as part of our mission is to deploy large language models safely, in direct contact with affected communities.”