OpenAI debuts 'imperfect' tool to catch ChatGPT-generated academic cheating, misinformation

OpenAI's ingeniously human-like chat bot ChatGPT can write smart essays and even pass college entrance exams, putting it in the crosshairs of school districts and universities. On Tuesday, OpenAI rolled out a tool it hopes will help ease concerns that it has unleashed the greatest machine for cheating and academic dishonesty ever gifted to unscrupulous students. The new tool, AI Text Classifier, is designed to help educators identify content that was created by ChatGPT, not their students.

But the tool isn't foolproof, OpenAI warned in a blog post Tuesday. The AI Text Classifier's method for detecting AI-generated text "is imperfect and it will be wrong sometimes," OpenAI's Jan Leike told The Associated Press. "Because of that, it shouldn't be solely relied upon when making decisions." In fact, "we don't fundamentally know what kind of pattern it pays attention to, or how it works internally," Leike said. "There's really not much we could say at this point about how the classifier actually works."

The new tool analyzes entered text and labels it "likely" AI-generated, "very unlikely," and shades in between. OpenAI said its AI detector works best with longer texts — certainly more than 1,000 characters — written in English. Along with fighting academic dishonesty, the tool could be used to detect AI misinformation campaigns or chatbots trying to pass off as human, the company said.

Subscribe to The Week

Escape your echo chamber. Get the facts behind the news, plus analysis from multiple perspectives.

SUBSCRIBE & SAVE

Sign up for The Week's Free Newsletters

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

Sign up

School districts and other educational institutions are experimenting with how to proceed in a world where ChatGPT is not just a reality, but almost certainly a harbinger of more advanced AI wordsmiths to come. School districts in New York City, Los Angeles, and other big cities blocked ChatGPT by the end of 2022, but Seattle Public Schools decided to lift the ban and treat it like a teaching tool, district spokesman Tim Robinson told AP, adding, "We can't afford to ignore it."

"The initial reaction was 'OMG, how are we going to stem the tide of all the cheating that will happen with ChatGPT,'" but "this is the future," Devin Page, a technology specialist with Maryland's Calvert County Public School District, told AP. "I think we would be naïve if we were not aware of the dangers this tool poses, but we also would fail to serve our students if we ban them and us from using it for all its potential power."