Designing Learning in the Age of AI: What Actually Works

In the last conversation, we asked a slightly uncomfortable question:
If AI can do the work, what exactly are we assessing?
This is where most discussions politely end.
Because the next step is harder.

It requires schools to stop talking about the problem—and start redesigning the classroom.
So let’s get to it.

First Principle: You Cannot Out-Police AI

Let’s dispense with the obvious. You can block websites. You can redesign exam halls.
You can tighten invigilation. Students will still find a way.
The more energy we spend on control, the less we invest in something far more effective:
designing learning that cannot be meaningfully outsourced.

What That Looks Like in Practice

At Hayde Heritage Academy, this has meant a deliberate shift away from assignments that reward polished answers—and towards those that reveal thinking.
Not louder. Not more complicated. Just… smarter.

1. Scaffolded Assignments (Where Process Matters More Than the Final Answer)

The easiest thing for AI to produce is a finished product.
The hardest thing for it to replicate is a learning journey.
So we stopped asking for just the final submission.
In our Grade 8 classrooms (Session 2026):
• Students don’t submit a single essay—they submit idea notes, drafts, reflections, and revisions
• Teachers respond mid-process, not just at the end
• Students are asked to explain why they changed something—not just what they wrote
Result?
Even if AI is used, it becomes a tool within the process, not a shortcut to bypass it.
And more importantly, students start realising that thinking cannot be copy-pasted.

2. Personal Context = Non-Transferable Learning

AI works best when questions are generic.
So we stopped being generic.
Instead of:
“Write about urbanisation”
We ask:
• How has your town changed in the last 10 years?
• What do your parents or grandparents notice?
• Where do you see opportunity—and where do you see loss?
In one Grade 8 task, students interviewed family members about changing work patterns and linked that to industrialisation.
You can use AI to structure your answer.
You cannot outsource your grandmother.
That changes the game.

3. Multimodal Assessment (Because Not All Thinking Looks Like an Essay)

The traditional model assumes that if a student can write well, they understand well.
That assumption has always been shaky.
AI has just made it collapse faster.
So we diversified.
At Hayde Heritage:
• Students write, but they also speak, present, debate, and visualise
• A concept explained orally often reveals far more than a written answer
• Peer questioning exposes gaps that polished submissions hide
When a student has to defend an idea in real time,
there is no “generate answer” button.

4. AI as Support, Not Substitute

Let’s be clear—AI is not the villain here.
Used well, it can be a remarkably effective learning companion.
We’ve seen this in our Grade 8 mathematics classrooms:
• Students struggling with algebra used AI tools for step-by-step clarification at home
• They practised more—because help was available without hesitation or embarrassment
• Classroom time then shifted to application, discussion, and error analysis
One student who consistently avoided problem-solving began attempting more questions simply because the first barrier—fear of getting stuck—was removed.
That’s not dilution of learning.
That’s access.

5. The Flipped Possibility (If We’re Willing to Let Go)

There is also a more radical shift available—if schools are willing.
Students engage with explanations (including AI-supported ones) at home.
Classrooms become spaces for:
• Application
• Discussion
• Correction
• Extension
In other words,
school becomes the place where thinking happens—not where information is delivered.
It sounds idealistic.
It’s actually practical—if designed well.

6. The Catch (Because There Always Is One)

Before we get carried away, let’s acknowledge the obvious risks:
• Over-reliance on AI can weaken independent thinking
• Easy answers can discourage deep engagement
• The line between “help” and “replacement” can blur quickly
Which is precisely why the teacher’s role becomes more—not less—important.
Not as a gatekeeper.
But as a designer, observer, and challenger of thinking.

So What Actually Changes?

Not everything.
But the important things do.
• We stop rewarding only the final answer
• We start valuing the process
• We design for thinking, not just completion
• We accept that tools will evolve—and learning must evolve with them

 
A Slightly Uncomfortable Conclusion
If a task can be completed entirely by AI,
it is probably not worth assigning in its current form.
That’s not criticism.
That’s clarity.

Where This Leaves Us

At Hayde Heritage Academy, the shift is still in progress. It is not neat. It is not finished.
But it is intentional.
Because the goal is no longer to protect old systems.
It is to build better ones.
AI is not going away.
The question is whether our classrooms are willing to move with equal intelligence.