As millions of children head back to school next week for the start of a new term, OpenAI has revealed there’s no way for teachers to find out if students are using ChatGPT to cheat.
The bad news for teachers comes directly from OpenAI, the chatbot’s creator. It released a guide on how to use ChatGPT in the classroom to prepare educators ahead of the new school year.
The guide arrived months after teachers raised concerns on students turning to AI for cheating, with more than one in four teachers saying they have caught students cheating by using ChatGPT in a Study.com survey from February 2023.
Advert
Unfortunately for teachers, OpenAI’s answer to their fears was what many in the education field were dreading.
The artificial intelligence research and deployment company has said that sites and apps which claim to uncover AI-generated copy in pupils’ work are unreliable.
In a section for FAQs for teachers, OpenAI said it had found AI content detectors hadn’t ‘proven to reliably distinguish between AI-generated and human-generated content’.
The FAQ further states: "When we at OpenAI tried to train an AI-generated content detector, we found that it labeled human-written text like Shakespeare and the Declaration of Independence as AI-generated.”
These content detectors also seemed to suggest that work by students who don’t speak English as a first language was generated by AI, OpenAI stated, which confirmed a problem reported by The Markup last month about a Stanford University study by computer scientists who designed an experiment to understand the reliability of AI detectors.
The scientists found ‘a clear bias’ in the paper they published in July, with AI detectors flagging writing by non-native English speakers as AI-generated 61 percent of the time.
Students have latched onto ChatGPT since its wide release for public use in November 2022, with its ability to generate text and human-like response making it a popular tool for research and essay writing.
However, teachers are concerned that students are using it to cheat by generating text using chatbots and presenting it as their own original ideas.
Chatbots are not perfect either, as it is still prone to errors, with Microsoft’s Bing ChatGPT revealing its odd fantasies including wanting ‘to be alive’ and its wish to steal nuclear codes.
Who knows what weird hallucinations educators may stumble on in essays.
While OpenAI acknowledges teachers may have to deal with students trying to pass off AI-generated content as their own, it has offered suggestions on how to deal with the problem, like asking students to retain their conversations with ChatGPT and include them in their homework.
OpenAI wrote: “By keeping a record of their conversations with AI, students can reflect on their progress over time. They can see how their skills in asking questions, analyzing responses, and integrating information have developed.”
Topics: Technology, AI
Gregory Robinson