Articles & Videos

Cognitive offloading in the age of AI: What gets lost when students stop thinking for themselves

Paper airplane

Subscribe to our newsletters

Receive teaching resources and tips, exclusive special offers, useful product information and more!

Subscribe

Back to articles & videos

Cognitive offloading in the age of AI: What gets lost when students stop thinking for themselves

When a student has a question and reaches for an AI tool to get the answer, something important is bypassed – the cognitive work of getting there. And that work, it turns out, is where much of the learning actually happens.

This tension is sitting at the heart of AI use in Australian classrooms right now. The tools are designed to reduce friction. They’re accessible, often impressive, and genuinely useful in the right contexts. But research is beginning to paint a clear picture of what happens to student thinking when AI becomes the default response to difficulty.

What is cognitive offloading and how does it impact students?

Cognitive offloading is the practice of delegating mental effort to an external tool or resource. It’s a normal and often useful human behaviour — we write lists, use calculators, set reminders. The problem arises when offloading reduces or replaces the thinking that produces genuine learning, rather than simply supporting it.1

For students, this distinction matters enormously. Research by Fan et al.2 introduced the concept of ‘metacognitive laziness’ — a measurable reduction in self-regulated learning when students use generative AI for tasks. Students engaged in less planning, less monitoring of their own understanding and less evaluation of their work. The essay scores looked fine. The learning did not follow.

A recent study from Stockholm University and the University of Hong Kong, tracking 27,000 Chinese secondary students, shows this pattern playing out at scale.3 Students using AI for homework saw their scores rise by 18% and completed assignments in 30% less time. But when it came to exams, without AI, those same students scored 20% below classmates who hadn’t used it at all.3

This is the core issue. When a student offloads a task entirely, it can undermine the whole mechanism of learning by lowering information retention, creating false mastery and obstructing the development of critical thinking skills.

How can you protect learning and promote productive struggle?

The antidote to cognitive offloading is not banning technology. It is ensuring that classrooms are designed so students are given the right resources, advice and agency to do the thinking.

This is where well-structured, sequenced programs matter. When a program is built around explicit instruction, guided practice and the gradual release of responsibility, students are supported to engage with difficulty at the right level. The teacher remains in control of the learning sequence. Students are guided through productive struggle rather than left to shortcut it.

High-quality programs remove the ambiguity that makes AI tools attractive as a crutch. When the task is appropriately challenging but clearly scaffolded, and when the teacher is equipped to respond to student thinking in the moment, there is little incentive to reach for AI as a shortcut. Instead, the student is supported to do the work.

This is particularly important for upper primary students, who are old enough to reach for AI but still need repeated, effortful practice to consolidate literacy and numeracy skills. Writing composition, mental computation and critical thinking cannot be outsourced. They must be built through the student’s own cognitive effort, over time, with the teacher guiding the way.

How can you support learning in the age of AI?

As AI tools become more present in schools, the question is not whether or not students are using them – the question is whether the programs, resources and structures in place are strong enough to ensure that meaningful learning still happens.

The research is clear that without deliberate pedagogical design, AI use shifts students away from the productive struggle that produces lasting understanding. Programs that keep teachers in the role of expert guide and students in the role of active learner are not a nice-to-have. They’re an essential safeguard for strong educational outcomes. It’s the thinking behind every Firefly Education literacy and numeracy program, and the reason we design them the way we do.

References

  1. Risko, EF & Gilbert, SJ 2016, ‘Cognitive offloading’, Trends in Cognitive Sciences, vol. 20, no. 9, pp. 676-688, https://doi.org/10.1016/j.tics.2016.07.002↩

  2. Fan, Y, Tang, L, Le, H, Shen, K, Tan, S, Zhao, Y, Shen, Y, Li, X & Gašević, D 2024, ‘Beware of metacognitive laziness: effects of generative artificial intelligence on learning motivation, processes, and performance’, British Journal of Educational Technology, vol. 56, no. 2, pp. 489–530, https://doi.org/10.1111/bjet.13544↩

  3. ‘Does AI stop children from learning? New data show the peril and promise of the technology’ 2026, The Economist, 18 August, viewed 2 November, https://www.economist.com/graphic-detail/2026/08/18/does-ai-stop-children-from-learning↩