Chatbot cheats face fines and, jail

Chatbot creators risk jail and stiff fines for ‘academic cheating’ if they commercialise artificial intelligence to write student essays and assignments.

Artificial intelligence can now write credible legal essays. Picture: iStock

Chatbot creators risk jail and stiff fines for “academic cheating’’ if they commercialise artificial intelligence to write student essays and assignments.

The Tertiary Education Quality and Standards Agency is evaluating whether ChatGPT, a nascent technology launched just weeks ago by the Microsoft-backed Open AI, is in breach of anti-cheating laws.

A spokeswoman said TEQSA would “identify risks and strategies’’ for dealing with artificial intelligence in academia this year.

“Where TEQSA identifies an AI service that may breach the prohibition on the provision of academic cheating services in the TEQSA Act, we will investigate and, where appropriate, take enforcement action,’’ she said.

Companies found to offer commercial essay-writing services, for use in university assignments or exams, risk fines of $687,500, with jail terms of up to two years for individuals. TEQSA has used its legal powers to force internet providers to block 152 websites offering commercial services to write essays and assignments.

But, while the blocked websites rely on human ghostwriters, ChatGPT is the first to use AI to generate comprehensive written responses on command.

The free website was never intended to be used for cheating, yet educators worldwide are scrambling to find ways to detect use and abuse of it in schools and universities as students embrace an easy way to generate individualised cut-and-paste answers, under the radar of plagiarism checks.

Law professor Stephen Colbran, Dean of Law at CQUniversity in Queensland, warns that ChatGPT can already write a credible assignment on Australian contract law and will only grow “smarter’’ with time. “We’re surprised by the quality of the answers coming out of it,’’ he said.

“The more information it gets, the smarter it gets, and it’s creative. We used to be worried about other people writing answers (for students), let alone a machine.’’

Professor Colbran said he did not advocate banning the bot, but would change his teaching and assessment methods.

“The challenge is for academics and teachers to assess students knowing the resource is there,’’ he said. “I will be asking students, ‘How did you come up with that? What sources of information did you find?’ We can have more group work, which students hate, or get them to do a presentation to explain their methodology, rather than spout out what the machine’s told them.

“Teaching will have to be more interactive and engaging.’’

Professor Colbran warned that the chatbot could also be “intelligently stupid’’ and generate well-written misinformation.

“It could learn the wrong stuff if someone gives it wrong information,’’ he said. “It doesn’t know right from wrong – it just analyses the information it has.’’

The TEQSA Act defines an “academic cheating service’’ as one that provides work to, or undertakes work for, tertiary students on a commercial basis, as a substantial part of an assessment task they are required to undertake themselves.

“If they (chatbots) provide systems that allow work to be completed for an Australian tertiary student, it could fall under that Act,’’ Professor Colbran said.

ChatGPT is free as a trial version but Open AI has announced it may “monetise’’ chatbot with a faster and more comprehensive “pro’’ version to offset “eye-watering’’ computing costs.

Education departments are now collaborating on ways to detect use of the essay-writing chatbot, after billionaire Elon Musk declared “Goodbye homework!”.

Open AI has been contacted for comment.