5:24 am today

Why AI won't ruin children's education

5:24 am today
Kid boy sleeping with a textbook over his head and holding a sign with the word Help

Photo: 123RF

A new version of Chat GPT helps students bypass detectors and cheat on their homework, but one principal says banning AI won't fix the issue.

New versions of Artificial Intelligence are changing the education system and helping students get away with cheating.

Some schools and universities have gone back to classic pen and paper exams and hand-written assignments, and have scrapped online tests after teachers noticed an increase in cheating using AI technology.

Some places, like parts of Australia, have responded to the issue by banning Chat GPT completely.

But Secondary Principals Association president Vaughan Couillault thinks a black and white solution isn't the answer, because the range of technology available and what it can do is incredibly broad.

"It's a sensational search engine that does a bit of summarising for you really," he says.

"There's the good side and then there's the darker side where people might be using generative AI to make plagiarism easier... and then there's everything in between."

A new report calling for a science-based approach to using AI in classrooms raised concerns about how using it too much before learning the basics may slow down a student's learning.

Couillault shares this concern.

"It's a walk before you can run thing, and when you think back to the evolution of how we do maths, where we used to do maths with bits of paper and then it moved to abacus and then it moved to calculator - through all that evolution we said, 'actually we need you to know how the thinking works around that in order to have an understanding later on and be able to calculate it," he says.

Auckland University student Jacob Capes studies mechatronics engineering and is currently preparing for his end of year exams, with the help of Chat GPT.

"I will take a screenshot of a [lecture] slide and copy and paste it into chat GPT. Then I can ask it to extrapolate what is on the slide in a way that I can understand it," he says.

Less than 10 seconds later, the virtual assistant will reply with several detailed bullet points worth of information. Capes says he can then go through the text, highlight the parts he still doesn't understand and get Chat GPT to explain further.

"This is really valuable for me because in a lecture you can't really ask a lecturer to keep going why, why, why, until it gets down to a root cause, whereas with Chat GPT I can just continually keep prompting it until I have an understanding, it doesn't matter how long it takes," he says.

While it's not using AI as a study tool, plagiarism and authenticity of work is the main concern for educators internationally.

Capes says he hasn't ever used AI to do his work for him, and that it's obvious when something has been written by a robot.

"It's almost too perfect and beyond what a normal human would sound like in a conversation or in writing... and comparing it to other Chat GPT written articles they all have very obvious similarities," he says.

But it appears technology developers are catching on, launching creations like StealthGPT, designed specifically to "humanise" writing and bypass AI detectors.

Couillault doesn't think AI is entirely to blame for cheating.

"Let's be honest, people have been writing the answers on their forearm under their sleeve for generations... basically what generative AI has done is made it hyper-easy," he says.

He says it is also relatively simple to identify if a student has used it in this way.

"For example, if a teacher knows a student pretty well they'll understand that that student always uses an 's' to spell organisation, but then suddenly in their essay all of their organisations had 'z's in them, [it] probably wasn't them writing it," he says.

So Couillault does not think a ban would be justified.

"What we find with new technologies like this is that the expectation of how good or bad it's going to be in five or 10 years' time is often quite different from how it actually lands," he says.

Couillault says the field of AI is both exciting and terrifying, which is why a measured approach to how it's used is essential.

"That's not saying don't do it, that's saying know what we're doing before we flick that switch because you can't un-ring a bell."

Check out how to listen to and follow The Detail here.

You can also stay up-to-date by liking us on Facebook or following us on Twitter.