The first problem
The first problem is that us humans are not very good at being able to detect whether student work is actually their own or if it is AI-generated. When asked by TeacherTapp in 2023 whether they suspected students were using AI without permission, only 6% of teachers said “Yes, on multiple occasions”. Fast forward twelve months and this number increases to 20%.
That may be what teachers suspect, but the reality is that probably even more are doing so. Many more.
The second problem
Problem number 1 wouldn’t be such a big issue if it wasn’t for Problem 2.
Problem number 2 is that, currently, AI detecting software isn’t very good. Researchers have found that the best AI detection technology is accurate on average around 67% of the time. This means that a lot of students are either a) wrongly accused or b) getting away with it. The researchers could not be clearer when they state that “detection tools are neither accurate nor reliable”.
