Education researchers are warning that schools adopting AI tools may be moving faster than they can fully understand the consequences.
A new joint study by the University of Technology Sydney (UTS) and the University of Queensland argues that AI can deepen learning – but only if governments move quickly to adopt draft national standards for safe, educationally sound tools and equip teachers to guide their use.
At stake, say co‑authors Professor Leslie Loble AM and Professor Jason Lodge, are the mental habits that sit at the heart of learning itself: concentration, memory, reasoning and the ability to think deeply through complex problems.
The risk of outsourcing young minds
The report, released by the Australian Network for Quality Digital Education, investigates the risk that students are outsourcing too much of the thinking needed to build knowledge, skills and lifelong learning capacity.
Professor Loble, who is the Network’s Chair and an Industry Professor at the UTS Centre for Social Justice & Inclusion, said AI brought with it a significant risk for students who outsource too much cognitive work crucial to establishing their knowledge, skills and “thinking infrastructure”.
“The cognitive offloading from human to AI is especially risky for school students, who are building the foundational knowledge and skills that enable both schooling and lifelong capacity for learning and understanding,” Professor Loble said.
Professor Loble, who is also former chair of the Australian Education Ministerial Council’s Schooling Policy Group and former Deputy Secretary in the NSW Department of Education, said the educational imperative is not to protect students from a world where AI is the norm, but to prepare them for it.
“Young people need strong knowledge foundations and complex reasoning skills – not just the ability to ‘Google it’ or outsource the thinking to AI.”
Professor Loble said the fact that students are already using AI extensively means this policy challenge “cannot be put in the too hard basket”, as was recently done with social media.
“The issue isn’t whether AI exists in classrooms, but whether it is being used to strengthen learning and help students become more effective thinkers,” she said.
“The decisions we make now will determine whether AI deepens students’ knowledge and critical thinking, or instead hollows out the learning process and causes long-term harm to their cognitive development.”
Professor Loble said AI’s propensity for error and hallucination makes it even more essential that students build deep knowledge and strong analytical skills.
“Teachers are a crucial part of the solution.”
The report says getting AI right in schools comes down to two things: smarter tools and stronger teaching.
The researchers point out that rather than simply spitting out answers, AI platforms should push students to think more deeply, stay engaged and build the foundational knowledge that underpins real learning.
Just as importantly, teachers need clear guidance, practical strategies and evidence-based support to use AI well in the classroom.
‘We risk creating a learning divide that will be very hard to close’
Cognitive psychologist Professor Lodge said a growing body of evidence showed that using AI can introduce serious risks for school students.
“School years are critical for building the memory stores and cognitive foundations that last a lifetime,” he said. “If we allow AI to replace that process for some students, we risk creating a learning divide that will be very hard to close. Further investment in research to understand these mechanisms is crucial.
Professor Lodge pointed out that while unstructured use of AI risks cognitive atrophy, humans still learn more effectively from and with other humans.
“By supporting the teacher, we empower the human expert who is best placed to manage the complex, relational work of co-regulating learning, managing cognitive load, and building the evaluative judgement, self-regulated learning, and metacognition that students need,” he said.
“Additionally, AI will almost certainly widen existing equity divides and mean a widening learning gap for disadvantaged students and schools if left unstructured.”
Professor Lodge said students with strong knowledge and learning skills will likely use AI to accelerate their learning, while disadvantaged students risk falling further behind by outsourcing the very thinking they need to develop.
“This further underlines the importance of giving support to teachers.”