If you walked down any school corridor today, you would notice a curious shift.
Where pupils once paused to wrestle with a tough essay question or debate a conflicting history source, you now see a quick series of taps on a smartphone screen. An unthinking prompt goes into ChatGPT or Claude, returning an answer three seconds later. On paper, homework is submitted faster and looks more polished, but genuine understanding and critical thinking fall short.
And that becomes obvious the next morning when you ask a simple follow-up question requiring them to connect those ideas, you get complete silence.
As educators, we are witnessing a phenomenon that researchers call harmful cognitive offloading. AI has not just made information accessible, but it has created an illusion of competence. By removing the natural friction of learning (the mental struggle), we are inadvertently robbing young minds of what cognitive psychologist Robert Bjork calls desirable difficulties: the intentional effort required to build deep, durable understanding.
To fix this, we do not need more complex technology or AI-detection software in our lesson plans. We need intentional simplicity and deliberate unpredictability.
The AI Offloading Trap: Fluency Without Mastery
When AI acts as an instant answer oracle, students completely bypass the generation effect: the proven cognitive principle that memory and understanding are significantly stronger when we generate answers ourselves rather than simply reading them.
-
The Passive Trap: Reading a beautifully structured AI-generated response feels satisfying, but it creates zero neural connections.
-
The Illusion of Effort: Pupils confuse the speed of getting an answer with the process of understanding it.
-
Erosion of Retrieval: When pupils rely on external tools for basic synthesis, their ability to retrieve and connect facts independently degrades over time.
"Technology saves time, but learning requires time. Technology reduces effort, but learning requires effort."
If every question has an immediate, predictable, algorithmic answer, learning becomes a sterile transaction.
Disrupting Predictability: The Power of Randomness
So how do we break the passive spell and reignite real, unconventional thinking?
We bring back surprise.
While structured routines in the classroom create emotional safety and lower unnecessary administrative distraction, intellectual novelty triggers attention. When pupils know exactly how every task will unfold, their brains shift into energy-saving autopilot. Introducing simple elements of randomness breaks that pattern and forces active engagement.
Here is how simple, unpredictable structures bring productive struggle back to the classroom:
1. Unpredictable Targeted Questioning
When you ask a question and call for hands up, only three eager student respond while twenty-five disengage. Instead, use a low-tech, high-engagement digital
When targeted questioning is genuinely randomised and lighthearted, every student stays mentally active (not out of anxiety, but out of playful anticipation). It turns passive listeners into active participants who are ready to offer an idea, critique a classmate's point, or summarise the discussion.
2. Random Prompt Mashups
Instead of handing out standard essay titles ('Analyse the causes of the First World War'), inject forced randomness. Have pupils roll dice or pick cards to combine two unrelated concepts:
-
'Analyse the French Revolution through the lens of modern social media algorithms.'
-
'Explain cellular respiration as if you were a detective writing a 1940s film-noir monologue.'
Unpredictable pairings render generic AI prompts useless because they require novel connection-making, which is the true hallmark of higher-order thinking.
3. Socratic Curveballs
Midway through a comfortable classroom discussion, throw in a contradictory piece of source material or an extreme edge-case scenario. Force pupils to stop, verify, and re-evaluate their assumptions rather than accepting the consensus.
Returning to Productive Struggle
| Algorithmic Learning (AI-Driven) | Unpredictable Active Learning |
| Goal: Get the answer instantly | Goal: Navigate the process of discovery |
| Mindset: Passive consumption | Mindset: Active inquiry and generation |
| Structure: Predictable linear output | Structure: Dynamic, surprising, non-linear |
| Result: Fragile, short-term recall | Result: Deep, durable mental models |
As teachers, our job in the age of generative AI is not to ban technology or build bigger digital walls. It is to make the physical classroom a space where answers are not handed out on a plate.
By embracing simple tools, randomised structures, and unconventional prompts, we reintroduce the joy of curiosity. Learning becomes exciting again not because it is easy, but because nobody in the room (including the teacher) knows exactly where the next twist will take us.
- Bjork, E. L., & Bjork, R. A. (2011). Making Things Hard on Yourself, But in a Good Way: Creating Desirable Difficulties to Enhance Learning. In M. A. Gernsbacher et al. (Eds.), Psychology and the Real World: Essays Illustrating Fundamental Contributions to Society. UCLA. Available at: https://bjorklab.psych.ucla.edu/wp-content/uploads/sites/13/2016/04/EBjork_RBjork_2011.pdf
- RM Compare. (2025). Cognitive Offloading Is Real, and New Learners Are Most at Risk: How RM Compare Keeps the Thinking Human. Available at: https://compare.rm.com/blog/2025/09/cognitive-offloading-is-realand-new-learners-are-most-at-risk-how-rm-compare-keeps-the-thinking-human/
- Wikipedia contributors. Novelty effect. Wikipedia. Available at: https://en.wikipedia.org/wiki/Novelty_effect
- Spin the Wheel. (2026). The Best Random Name Pickers for Teachers: Boost Participation Without the Anxiety.
