AI generated adaptive homework on the cards

A school in Queensland says its student performance tracking artificial intelligence system should be able to create ‘adaptive homework’ for individual students next year.

5/62024: Islamic College of Brisbane CEO Ali Kadri with students (L) Rahma Ibrahim, Year 10, and Ashratullah Walid Noori, Year 12. Picture: Lyndon Mechielsen / The Australian

A school in Queensland says its student performance tracking artificial intelligence system should be able to create “adaptive homework” for individual students next year.

The Islamic College of Brisbane has been beta-testing AI-powered student performance tracking for the last two months, CEO and principal Ali Kadri said. The model ingests a wide range of variables like student attendance, assessment performance, behavioural factors, teachers’ comments, or check-ins at the school’s wellbeing hub to generate a prediction for a student’s future grades.

Mr Kadri said this helped the school and parents intervene early if a student’s academic performance showed signs of dipping, as well as provide tailored career advice.

“Students, during their life cycle in a school, their academic results go up and down,” he said. “That’s often picked up after the fact. For example, you’ve got a student who’s doing really well in education until like Yr 9, and suddenly they start dropping.

“But by the time you realise, it’s really too late to put in any strategy to address those kinds of drops. So if you get indications that’s going to happen, you can put in early intervention strategies.

“And you can assist the student – maybe it’s an academic issue, they don’t understand; maybe it’s something to do with social behaviour or a wellbeing issue. And you can only do that if you intervene earlier.”

Mr Kadri said the data were not attached to students’ names for privacy reasons.

“Adaptive homework, that’s the future of the system,” he said. “Obviously, as more and more enhancements happen in AI – right now, theoretically that is possible, but the language models have not been tested enough to do this at this moment in time.

“Imagine a world where, for example, two students are in the same classroom. And they’re both studying Year 3 English. Now the gaps student A has are different to the gaps student B has.

“For a teacher to make that assessment and adapt the homework, means if she’s got 26 students, she has to write 26 assessments, which is impossible for any human being.

“Now if you’ve got a model that’s been trained in both of their education journeys and know where their gaps are, then that model can prepare an assessment, get the teacher to verify and authorise that assessment, and then issue that assessment.

“With our trajectory, we should be able to have something in place next year.”

The Islamic College of Brisbane has been an early adopter of technology recently and was one of the first in the country to bring ChatGPT into the classroom.

Mr Kadri said the new AI system had already helped with early intervention for a student whose grades were predicted to drop.

He said that the consistently high scoring student had seen a drop in their attendance but not enough to trigger an automatic response in the school system.

“Now this student was in high 90s for attendance but along with that, there were some behavioural incidents that were never noticed with this student previously, plus some voluntary visitations to our wellbeing hub,” he said.

“The data were already there – it’s the analysis of it and the novel way of analysing it, that’s what the AI is doing.”

He said that the school would not have been able to pick this up as quickly without the AI-powered system.

“Because the students in high school have different teachers for different subjects, one teacher would not be able to see all this,” he said.

“And it’s not resulted in a negative academic outcome yet, but it will, according to the algorithm.

“We would have intervened in this case, but after the fact. We only would have been able to intervene once their results had dropped.”