Software Schools Use to Detect Cheating Is Flagging Real Essays as AI Generated

A whopping 2.1 million teachers in the US are using a new AI detection tool by Turnitin, the Washington Post reports, in an attempt to catch their pupils in the act of using tools like ChatGPT to cheat in class. But as it turns out, the tool isn’t very good at what it was designed to do, which is likely resulting in students being wrongfully accused of having used AI tools for essays and assignments — a worrying side effect of putting these tools in the hands of practically anybody with an internet connection. The reality is that while AI chatbots have exploded onto the scene and are continuing to improve, tools that are capable of distinguishing between AI-generated and human-written text are woefully behind. In other words, it’s a huge problem without a solution that is putting both teachers and students in a precarious position. WaPo’s Geoffrey Fowler gave Turnitin’s AI detection tool a whirl by testing it with the help of five high school students. Out of 16 samples, which included human-written, AI-generated, and mixed-source essays, the tool was wrong more than half the time. In short, the tool probably shouldn’t be used by teachers to accuse their students of using AI, despite Turnitin claiming that its detector is 98 percent accurate. In all fairness, the company does point out on its website that the results of its tool shouldn’t be used to accuse students of cheating, but whether that’s going to stop every teacher out there seems unlikely….Software Schools Use to Detect Cheating Is Flagging Real Essays as AI Generated

Leave a Reply

Your email address will not be published. Required fields are marked *