This morning, The Guardian ran a great story on an Australian university lecturer detecting AI bot use in one-fifth of assessments, raising concerns for AI use in exams. Essentially, AI text generation has been helping university students cheat.
The Guardian’s report is the latest article covering the controversies of AI text generation bots, with a similar story breaking in the U.S., bots using AI text tools on social media and, of course, scams. Similarly, AI-generated art has garnered massive controversy.
“The big issue is how do educators revise assessments in light of this technology – which is not going away – so that we can fairly and accurately assess student competencies,” Deacon University associate communications lecturer Sally Brandon told The Guardian.
“I’m looking at how it could be incorporated in learning design and how assessment needs to be revised to account for its use.”
So, obviously, it’s a terrible idea to cheat on an exam. The results for getting caught could be catastrophic to your study, and even if you do make it through, you’re unable to absorb the information you’re being assessed on (which may come back to sting you later in your career).
But to make matters worse? AI’s are limited in their depth of knowledge.
The problem with using AI to cheat
It’s certainly worrisome that university students can cheat on assignments with the help of an AI. Unfortunately too, it’s not at all unbelievable, considering how sophisticated the ChatGPT app appears to be.
If you’ve been out of the loop, ChatGPT is an AI that will generate you text depending on what prompt you type in. It accesses a gigantic body of reference material to find the information that best correlates with your prompt.
The results, while mixed, have generated some realistic bodies of text.
No further questions #ChatGPT pic.twitter.com/JXfqBrfEH6
— Tim Burrowes – Unmade (@timburrowes) January 16, 2023
So, hypothetically, if you were to plug in a prompt similar to (or exactly the same as) the question on your essay, you might be able to get the AI to generate the work for you.
And to make matters more complicated, and as The Guardian reported, technology that attempts to detect the use of AI-generated text has had mixed results.
But of course, the AI isn’t perfect. The ChatGPT website indicates that the bot is more about having an articulate dialogue than getting information right (as echoed in The Guardian’s report by computer scientist and Curtin University associate professor Nik Thompson).
Very interesting response by #ChatGPT it just made up a fake Jimmy Carter book about Omar Torrijos @OpenAI pic.twitter.com/gOVA6vQVCG
— 𝙲𝚢𝚛𝚞𝚜 𝙰𝚜𝚑𝚊𝚢𝚎𝚛𝚒 (@cyrusashayeri) January 16, 2023
With this in mind, although your AI-generated essay might look feasible at first glance, the information and sources could be entirely made up.
The bot is only as correct as the datasets it trawls through. It’s likely that relevant information isn’t on its radar.
And if it pulls a piece of information word-for-word from a source? That’s plagiarism, baby. Especially if it isn’t credited.
Nobody has a problem with things like autocorrect or services like Grammarly, which try to improve your grammar and comprehension artificially, but relying on an AI for your entire paper is… Probably not going to go the way you think it is.
So, could you use AI to cheat in university? Yeah, totally. Would it be accurate? It depends on its ability to find correct information relative to the prompt. Should you use an AI to cheat? Absolutely not.