As students turn to AI to do their assignments and detection tools fail, universities scramble to rethink their assessment methods - while some institutional denial about the scale of the problem abides.
Listen to this article 14 min Listen to this article 14 min A crisis is brewing in local higher learning institutions over the use of artificial intelligence (AI) to cheat - and university authorities seem to be at sea over how to handle it.
Charne Lavery, associate professor in English at the University of Pretoria, first realised the scale of the problem when reading her 800 second-year students' essays two years ago.
The submissions shared an uncanny uniformity - flawless grammar and impeccable format, but with a synthetic flatness that set alarm bells ringing.
"These essays just sounded completely different to what we would get in the past: this bland tone with a perfect essay structure," says Lavery.
When run through a plagiarism detector that many universities rely on, Turnitin, most essays came back clean. Yet Lavery estimates that 70% to 80% of the essays were to some degree AI-generated.
The problem? "The burden of proof is on the academic. And there is really no way to prove it at all," she says.
South African universities have battled funding crises, student protests and pandemic disruptions over the past decade.
Now they face an insidious new challenge: across the country, students...
