Yet with the rise of generative AI, that assumption is increasingly fragile. When high-quality outputs can be produced with minimal human input, the traditional alignment between product and learning begins to break down.

The question is no longer simply “What did the student produce?”

It is becoming “What did the student do to produce it?”

Why effort and process now matter more than ever

At the heart of this shift is a construct that education has historically struggled to assess: effort.

Effort is not directly observable. It sits behind the scenes, entangled with:

• time spent,

• cognitive engagement,

• persistence through difficulty,

• and the strategies used to solve problems.

Research has consistently shown that performance reflects a combination of ability and effort, yet our assessment systems rarely distinguish between the two. A polished submission may signal deep understanding, or it may reflect tool-assisted production with minimal engagement.

In an AI-enabled environment, this ambiguity is amplified.

If assessment remains product-focused, we risk evaluating:

• tool proficiency over learning,

• output quality over intellectual engagement,

• and in some cases, automation over authorship.

The limits of traditional assessment

Traditional assessment methods assume a relatively stable relationship between:

effort → learning → product

AI disrupts this chain.

Students can now:

• generate essays, code, or reports rapidly,

• iterate at a scale previously impossible,

• bypass much of the cognitive struggle that supports learning.

This does not make AI inherently problematic. It simply means that products alone no longer provide sufficient evidence of learning.

What does it mean to assess process?

A process-oriented approach shifts attention from what was produced to how it was produced.

This includes examining:

1. Time and engagement

• How long did the student engage with the task?

• Was the work distributed over time or completed in a single burst?

Digital environments now allow us to capture:

• time-on-task,

• interaction logs,

• revision histories.

These are imperfect, but they offer insight into learning behaviours rather than just outcomes.

2. Iteration and development

Learning is rarely linear. It involves:

• false starts,

• revisions,

• refinements.

Assessing process means valuing:

• drafts,

• version histories,

• evidence of improvement.

A final submission becomes one point in a trajectory, not the sole object of evaluation.

3. Cognitive effort

Effort is not just time, it is mental work.

Indicators include:

• tackling challenging problems,

• engaging with feedback,

• demonstrating conceptual change.

Cognitive load theory reminds us that meaningful learning requires investment of mental effort, not avoidance of it.

4. Decision-making and justification

Students should be able to explain:

• why they made particular choices,

• how they evaluated alternatives,

• where they used tools, including AI, and why.

This moves assessment towards thinking, not just producing.

Designing for visible effort

One of the most important insights from the research is this:

Effort is best assessed when it is made visible through design.

Rather than trying to detect effort after the fact, we can structure tasks so that effort leaves a trace.

Practical strategies include:

Staged submissions

Break tasks into components:

• proposal,

• draft,

• feedback response,

• final submission.

Each stage captures part of the process.

Process artefacts

Require students to submit:

• planning notes,

• annotated drafts,

• reflection statements,

• prompt logs when using AI.

This aligns with the principle that evidence of authorship is more robust than attempts at detection.

In-class or synchronous checkpoints

Short, controlled activities:

• discussions,

• oral explanations,

• quick problem-solving tasks.

These provide snapshots of unassisted thinking.

Reflective components

Ask students to articulate:

• what they found difficult,

• how they overcame challenges,

• what they would do differently.

Reflection reveals metacognition, a key component of effortful learning.

Rethinking grading: should effort count?

A common concern is whether effort should be graded directly.

The literature suggests caution.

Effort alone is not learning. However, process indicators can provide evidence of learning behaviours that lead to understanding.

A more balanced approach is to:

• assess quality of thinking and development,

• reward engagement with the process,

• and ensure that outcomes still matter.

In other words, effort is not a replacement for achievement, but a lens through which achievement is interpreted.

AI as a partner in process, not a shortcut to product

An interesting paradox emerges.

The same technologies that disrupt product-based assessment can also enhance process-based assessment.

AI tools can:

• support brainstorming and iteration,

• provide immediate feedback,

• help students explore alternatives.

The key is not to exclude AI, but to integrate it transparently into the learning process.

When students document:

• how they used AI,

• what they accepted or rejected,

• how they verified outputs,

they demonstrate critical engagement, not passive reliance.

A shift in mindset

Moving from product to process is not simply a change in assessment design. It is a shift in what we value.

From:

• correctness → thinking

• completion → development

• output → engagement

This aligns closely with long-standing educational principles, but AI has made the shift urgent rather than optional.

Concluding thought

If education is fundamentally about learning, then assessment must capture learning as it happens, not just what remains at the end.

In an AI-enabled world, the most meaningful question we can ask is not:

“Is this work original?”

but rather:

“What does this work reveal about the learner’s thinking, effort, and development?”

When we begin to answer that question with confidence, we move closer to assessing what truly matters.

I’ll admit it.

ChatGPT has become my new best friend.

Not in the “I’ve replaced human interaction” kind of way, but in the way every good collaborator earns their place. It’s there when I need to think something through, test an idea, or untangle a problem that would have taken me hours just a few years ago.

And yes… it makes some things ridiculously easy.

Coding? Faster.

Debugging? Less painful.

Working with Excel formulas that used to require a deep dive into forums? Now it’s a conversation.

Instead of searching, scrolling, and stitching together half-answers, I just ask.

And it responds.

The Illusion of Effortless AI

From the outside, it probably looks like magic.

Type a question AND get a polished answer.

Ask for code AND get something that works.

Need an Excel formula AND it’s done in seconds.

It’s tempting to think:

“This is easy. Anyone can do this.”

But here’s the part we don’t talk about enough…

It only looks easy because of what’s happening behind the scenes.

What 30 Years of Computer Science, Teaching and Dev Experience has Taught Me

I’ve spent over 30 years working in computer science and software development.

That matters.

Not because I can code faster than everyone else, but because I know how to think in this space.

I know:

• what a good answer looks like

• when something is slightly off

• how to refine a question to get closer to what I need

• how to break a messy problem into smaller parts

So when I “talk” to ChatGPT, I’m not just asking questions.

I’m iterating.

I’m probing.

I’m steering the conversation.

In many ways, I’m doing what I’ve always done as a developer just with a much more responsive partner.

Why It Feels Like a Superpower

This is where things get interesting.

AI hasn’t replaced my expertise.

It’s amplified it.

What used to take:

• 2 hours of debugging

• 30 minutes of Googling

• multiple failed attempts

…can now happen in minutes.

But the thinking hasn’t disappeared.

If anything, it’s become more important.

Because now the bottleneck isn’t:

“Can I find the answer?”

It’s:

“Can I ask the right question?”

And This Is Where Students Struggle

For many students, this is the gap.

They see the same tool I do, but they don’t get the same results.

Not because they’re less capable.

But because they haven’t yet built:

• the vocabulary

• the conceptual understanding

• the mental models

• the confidence to challenge or refine an answer

As I’ve explored throughout The AI Educator, every new technology exposes existing cracks in how we design learning.

AI is no different.

It doesn’t just give answers.

It reveals who knows how to engage with answers.

AI Isn’t the Shortcut. It’s the Mirror

Here’s the uncomfortable truth:

AI doesn’t level the playing field.

It often widens the gap.

Those who know how to think, question, and refine?

They get exponentially better results.

Those who don’t?

They get fast answers… but shallow understanding.

As highlighted in my book, when tools do the thinking for us, we risk bypassing the very processes that lead to deep learning.

So What Do We Do About It?

If ChatGPT is going to be everyone’s “best friend,” then we need to teach people how to have better conversations with it.

That means helping students:

• build domain knowledge (so they can recognise good vs bad outputs)

• develop vocabulary (so they can ask precise questions)

• practise prompting as a form of thinking; not just asking

• reflect on AI responses, not just accept them

Because prompting isn’t a trick.

It’s a literacy.

In fact, as I’ve argued elsewhere, it’s quickly becoming one of the most important skills students can develop, right alongside reading, writing, and critical thinking.

Final Thought

ChatGPT might be my new best friend.

But like any good friendship, it works because of what I bring to the table.

The better I think, the better it responds.

And that’s the shift we need to help students understand:

AI doesn’t replace your intelligence.

It reveals it.

You point out the mistake.

You ask for evidence.

You explain why the answer is wrong.

Me:

Can you provide the citation for the study you mentioned?

AI:

Certainly. Smith & Johansson (2018), Journal of Advanced Educational Cognition.

Me:

That journal doesn’t exist.

AI:

Thank you for p…

Read more

The campus is quieter, inboxes are filling with out-of-office replies, and everyone’s pretending they’re not already mentally on the couch finally getting stuck into Stranger Things Series 5. It’s the perfect time for reflection, not the heavy, “rewrite your teaching philosophy” kind, but the gentler end-of-year stocktake.

So, in the spirit of the season, I’ve been thinking about AI in education… Christmas-style.

The Ghost of Technologies Past

Every December feels a bit like A Christmas Carol for educators.

Remember when calculators were going to “ruin maths”?

When Google would destroy memory?

When Wikipedia would end scholarship forever?

Sound familiar?

AI is just the latest visitor knocking on the classroom door, and like every guest before it, it’s revealing more about our systems than the technology itself. When assessment breaks under AI, that’s not AI being naughty, it’s a signal that something needed rethinking anyway.

That’s not a failure. That’s feedback.

Naughty or Nice? (Hint: That’s the Wrong Question)

All year I’ve been hearing:

• “Students are using AI to cheat”

• “We need better detection”

• “We should just ban it until January”

I get it. Exhaustion lowers everyone’s tolerance.

But framing AI as naughty or nice misses the point. Tools don’t have intentions. They expose design choices. If a task can be completed without thinking, AI will happily do it and that tells us something important about the task.

Christmas isn’t about catching people out and neither is learning.

Stuffing the Pedagogical Stocking

If I could slip a few things into educators’ stockings this year, they’d be small but mighty:

• Clear permission, not silence

  Students do better when expectations are explicit. Ambiguity breeds anxiety (and shortcuts).

• Process over polish

  Drafts, reflections, decisions, and missteps, these are harder to fake and far more educational.

• AI as a thinking partner, not Santa’s workshop

  Let students argue with it, critique it, and improve on it. That’s where learning lives.

None of this requires a complete curriculum rewrite. Often, it’s just a shift in emphasis.

A Quiet January Reset

The best thing about Christmas isn’t the presents. It’s the pause.

January gives us permission to ask:

• What do I actually want students to learn here?

• Where does AI help thinking and where does it replace it?

• What’s one small change I could trial next term?

Not a revolution. Just a reset.

Final Thought (Before We All Log Off)

AI isn’t the Grinch stealing learning outcomes. It’s more like a very honest mirror. Sometimes what it reflects is uncomfortable but it’s also incredibly useful.

So if this year felt messy, uncertain, or unfinished… that’s okay. Education has always evolved in moments like this.

For now, close the laptop. Step away from the LMS. Let the questions sit quietly until the new year.

I’ll see you in January with fresh energy, clearer intent, and probably the same coffee.

Happy holidays 💫

Then we graded the final answer.

Now AI has walked into the room and quietly asked a confronting question:

If a machine can produce the product, what exactly are we measuring?

This isn’t an AI problem. It’s an assessment problem.

The uncomfortable truth about “good” assessment

Most traditional assessments reward polish.

• A clean essay.

• A correct solution.

• A finished project.

But those artefacts have always been proxies for learning, not learning itself. We’ve just been willing to accept the shortcut because, until recently, producing the product usually required going through the process.

AI breaks that assumption.

A student can now generate a plausible essay without wrestling with ideas. They can submit code they don’t understand. They can hand in work that looks like learning without ever engaging in it.

And when that happens, banning tools or chasing detection misses the point. The issue isn’t that students are skipping the process. It’s that our assessments let them.

Why process has always mattered (even when we ignored it)

Anyone who’s taught for more than five minutes knows where learning actually happens:

• In false starts

• In revisions

• In half-formed ideas

• In feedback conversations

• In moments of confusion that eventually turn into clarity

That cognitive journey is where understanding is built. The final product is just the residue.

Historically, education valued this. Oral exams. Drafts. Studio critiques. Apprenticeships. Dialogue. Iteration.

We moved away from that not because it was better pedagogy, but because it was easier to scale, standardise, and mark.

When assessment rewards outcomes, students optimise for outcomes

Here’s the part we don’t always like admitting: students are incredibly rational.

If the system rewards the final answer, they will focus on the final answer.

If efficiency is valued over effort, efficiency wins.

That’s not cheating. That’s alignment.

So when AI offers a faster route to the same grade, it shouldn’t surprise us that students take it. The real question is: What are we signalling that we value?

If the answer is “a polished submission,” AI will happily deliver that.

Designing assessment that makes thinking unavoidable

Measuring process doesn’t mean adding busywork or surveillance. It means designing tasks where thinking leaves fingerprints.

Some practical shifts that matter:

• Drafts with purpose

  Not “submit three drafts,” but drafts that show decision-making, revision, and response to feedback.

• Process artefacts

  Design logs, reflection notes, prompt summaries, version histories. Not as add-ons, but as assessable evidence.

• Checkpoint conversations

  Short vivas, critiques, or discussions where students explain why they made certain choices.

• Transparency over policing

  Clear expectations about AI use, paired with accountability for reasoning, verification, and judgment.

In other words, stop trying to prove whether AI was used. Start asking students to show how learning happened.

AI can support process if we let it

Here’s the twist: AI doesn’t have to be the enemy of process-based assessment. In fact, it can make it stronger.

AI can:

• Support brainstorming without replacing judgment

• Offer feedback that students must evaluate and refine

• Surface alternative perspectives students have to critique

• Make iteration faster, not thinking optional

But only if we design for that.

When assessment rewards reflection, explanation, and decision-making, AI becomes a thinking partner, not a shortcut.

This is a values question, not a technology question

At its core, measuring process over product forces us to answer a deeper question:

Do we care more about what students submit, or about who they are becoming as thinkers?

AI has simply removed the illusion that those two things are the same.

If we want learning to matter, our assessments have to make learning visible. Not just at the end, but all the way through.

Because the future of education won’t be defined by whether students use AI.

It will be defined by whether we still know how to recognise learning when we see it.

Not terrified of robots taking their jobs, but terrified of getting AI wrong.
Terrified of accidentally lowering learning standards.
Terrified of letting something slip through the cracks.
Terrified of being judged for even trying to use AI.

Underneath all that fear was a simple truth:
Educators want to do the right thing. They just don’t want to break anything in the process.

My goal with the book was to help relieve that fear. To give teachers a clear, calm, grounded starting point and show them that they can use AI confidently, ethically, and meaningfully without feeling overwhelmed or exposed. AI isn’t something to fear; it’s something to understand, to work with, and to shape with intention.

But reassurance doesn’t mean pretending the risks don’t exist.

Confidence with AI requires clarity about its pitfalls.
And when we’re designing curriculum and assessment, we need to stay alert to the places where AI can quietly undermine learning if we’re not paying attention.

That’s why a recent systematic review on the harms of generative AI in computing education caught my attention. Not because it contradicts the message of The AI Educator, but because it reinforces it. The paper, Beyond the Benefits: A Systematic Review of the Harms and Consequences of Generative AI in Computing Education. (Bernstein, S., et al. 2025) names some of the hidden, structural risks we’ve been sensing for a while now, and its findings align closely with the themes woven throughout the book: process over product, epistemic integrity, and the fragility of traditional assessment in an AI-rich world.

Below, I break down what the review found and why these insights matter so much for curriculum design going forward.

1. AI Is Making Learning Look Better Than It Actually Is

One of the boldest claims in the review is also the simplest:

Students can now produce work that looks competent without actually being competent.

This is the core of the new learning-loss problem. Not post-pandemic, but post-GenAI.

When a tool can generate polished code, tidy up conceptual explanations, or reorganise an essay into something coherent, it creates a convincing surface that masks the underlying gaps. It’s not cheating; it’s a quiet erosion of the learning process.

In The AI Educator, I talk about how education has always wrestled with fears surrounding new technologies, from calculators to The Internet, but GenAI introduces something distinct. It allows students to bypass the entire cognitive journey of learning, not just parts of it.

“When assessment focuses primarily on the end product… GenAI can act as a shortcut that bypasses understanding entirely.” Chapter 5, The AI Educator

The work looks fine. The learning doesn’t.

That’s the danger.

2. Cognitive Offloading Is Turning Into Cognitive Dependency

Every tool encourages some level of cognitive offloading. Writing did. Google did.

But GenAI accelerates it in ways that are pedagogically disruptive.

Students in the review reported relying heavily on AI for:

• debugging

• problem-solving

• generating ideas

• structuring arguments

• clarifying concepts

And the more they relied on AI, the more their confidence dropped.

The danger isn’t that students get help. It’s that they stop trying before they start.

“What was once an opportunity to wrestle with complex ideas becomes a button-click exercise in outsourcing thought.” Chapter 6, The AI Educator.

3. Students Are Losing Motivation, Not Just Skills

This is where the review gets especially interesting.

Students weren’t just losing knowledge. They were losing:

• agency

• ownership

• pride

• clarity about what they actually understood

AI didn’t make learning easier. It made learning feel less theirs.

If students feel disconnected from their own thinking, then no amount of polished output will matter.

“AI may generate content, but it cannot replace the human work of interpretation, empathy, and relational care.” Chapter 8, The AI Educator.

4. Hallucinations Become Misconceptions That Stick

One of the clearest risks in the paper is the problem of “false fluency.”

Generative AI often produces outputs that:

• sound right

• look right

• feel right

…but are structurally or conceptually wrong.

And novices, who are the heaviest users, are the least able to spot those errors.

AI doesn’t just hide gaps.
It creates new ones.

The risk is not exposure to bad information; it’s the illusion of accuracy.

“AI outputs often sound good but are structurally incorrect or conceptually flawed, embedding new misconceptions rather than resolving existing ones.” Chapter 6, The AI Educator

5. Traditional Assessment Is No Longer Fit for Purpose

The review reinforces something we’ve already been feeling: assessment cannot continue as usual.

Tasks that rely on:

• take-home writing

• unguided problem-solving

• surface-level explanation

• open-ended homework
  …are now vulnerable by design.

The problem isn’t student misconduct.
It’s misalignment.
The task asks for something AI can do better than a beginner.

That’s a design flaw, not a moral one.

“Traditional assessment models… may inadvertently encourage shortcuts when students perceive assignments as mere hurdles rather than meaningful learning experiences.” Chapter 3, The AI Educator

“Assessment becomes performative rather than diagnostic.” Chapter 5, The AI Educator

6. The Equity Issue That’s Easy to Miss

Finally, the review points to a subtle but important pattern:

Students with strong self-regulation, critical thinking skills, or access to human support tend to use AI as a thinking partner. A point I made in my last blog post.

Students without those advantages use AI as a replacement.

And that widens gaps we are already struggling with.

Chapters 2 and 3 of The AI Educator both highlight this tension:

Without guardrails, AI shifts from tool to tutor and not always in ways that preserve equity, agency, or epistemic integrity.

The more powerful the technology, the more powerful the disparity.

So What Do We Do With This?

Here’s the hopeful part. The review doesn’t argue that AI harms learning.
It argues that poorly aligned pedagogy harms learning in the presence of AI.

And that’s a crucial distinction.

Across the book, one message keeps resurfacing:

• AI is not the threat.

• Misaligned curriculum is the threat.

• Product-focused assessment is the threat.

• Lack of guardrails is the threat.

• Uncritical adoption is the threat.

GenAI isn’t breaking education.
It’s revealing where it was already fragile.

And in that sense, it’s giving us an opportunity.

If we design for process over product…
If we make students show their thinking, not just their answers…
If we teach AI literacy rather than fear…
If we integrate guardrails instead of outsourcing judgment…

Then AI becomes a catalyst and not a shortcut.

A Practical, Forward-Facing Call to Action

If you’re an educator designing curriculum right now, here’s where I’d start:

1. Identify the thinking that AI cannot do.
That’s what your curriculum must foreground.

2. Shift assessments toward visible processes, not polished products.
Think logs, drafts, steps, reflections, decisions.

3. Normalise disclosure rather than detection.
Make AI use part of the assignment, not something to police.

4. Build AI literacy into the curriculum itself.
Students need to critique AI, not copy it.

5. Make human judgment, empathy, and dialogue central.
These are your competitive advantages and AI amplifies them.

Closing Thought

When I set out to write The AI Educator, my intention was simple:
to help educators use AI without fear.

This new research doesn’t undermine that goal.
It strengthens it.

Because to use AI confidently, we need clarity.
To design with AI responsibly, we need honesty.
And to build curriculum for an AI-rich world, we need to guard the parts of learning that matter most.

Not because AI threatens them.
But because they have always been worth protecting.

Reference

Bernstein, S., Rahman, A., Sharifi, N., Terbish, A., & MacNeil, S. (2025). Beyond the Benefits: A Systematic Review of the Harms and Consequences of Generative AI in Computing Education. arXiv preprint arXiv:2510.04443.

What they are skilled at is Googling. In fact, even that statement is questionable. For decades there’s been this myth that millennials and younger are some kind of “digital natives” and have an intuition for using tech. But this is not the case and that’s maybe a rant for another day.

Generative AI isn’t a search engine.
And treating it like one barely scratches its potential.

A Google-style query retrieves information.
An AI dialogue builds understanding.

When students interact with AI the same way they interact with Google, they miss the real power of these tools which is the power to critique, challenge, refine, simulate perspectives, test assumptions, and stretch their thinking.

Let me show you exactly what I mean.

Students aren’t misusing AI out of laziness, they’re misusing it because no one has taught them the difference between searching and thinking with a machine.

The Before/After: What Students Actually Ask vs. What They Should Ask

These examples illustrate the gap between search queries and learning dialogues. A gap educators can close with good scaffolding.

Example 1: Understanding a Concept

BEFORE (Google-style):
“Explain disruptive innovation.”

This produces a neat definition, but no thinking, no struggle, no conceptual development.

Why is struggle important to learning? Check out this post.

AFTER (dialogic, critical, Metis-style):
“Here’s what I think disruptive innovation means:
‘A new technology that replaces an old one by being cheaper or more convenient.’
Where is my explanation oversimplified? Ask me two questions that help me deepen it.”

Now the student is:

• articulating their prior understanding

• inviting challenge

• stepping into a dialogue

• refining their mental model

This is learning, not retrieval.

Example 2: Preparing for a Discussion or Workshop

BEFORE:
“What are the pros and cons of XR in higher education?”

Again, a list, not a learning moment.

AFTER:
“I’m preparing for a debate on whether XR is transformative in higher education.
Challenge my position that it’s transformative.
Give me a counterargument from the perspective of a university CFO.”

Now the AI becomes a sparring partner, not a list-maker.
The student practices argumentation, perspective-taking, and justification.

Example 3: Working With Ambiguity (instead of avoiding it)

BEFORE:
“Will quantum computing be disruptive?”

This frames AI as a prediction oracle.

AFTER:
“Give me two plausible interpretations of whether quantum computing is disruptive, and explain what assumptions each interpretation relies on.
Then ask me which interpretation I find more convincing and why.”

Students learn to:

• recognise ambiguity

• articulate criteria

• evaluate competing views

• confront uncertainty (instead of fleeing from it)

Example 4: Strengthening Reasoning (instead of bypassing it)

BEFORE:
“Is AI a sustaining innovation or a disruptive one?”

A fast question that gives a fast answer leading to shallow learning.

AFTER:
“Help me classify generative AI by asking me three diagnostic questions first.
Don’t offer your own definition until I’ve attempted one.
Then identify two assumptions in my reasoning.”

This mirrors the kind of structured Socratic inquiry that deep learning thrives on.

Students aren’t misusing AI out of laziness, they’re misusing it because no one has taught them the difference between searching and thinking with a machine.

This is where scaffolding becomes essential.

Why Scaffolding Responsible AI Use Is Non-Negotiable

Every semester, I see two kinds of students:

1. The ones who rush straight to polished AI-generated answers.

2. The ones who feel anxious and afraid of “using AI wrong.”

Neither group is learning deeply.

And without guidance, generative AI doesn’t just become a shortcut, it becomes a bypass around critical thinking, reflection, and productive struggle. These are the exact processes higher education is meant to cultivate.

Scaffolding AI use isn’t about restriction. It’s about helping students think better with these tools rather than think less.

A Light Touch of Metis: A Guide, Not a Crutch

In my own course, I introduce Metis, a fictional guide inspired by the Greek figure of strategic wisdom. Metis appears at the end of each topic to model how students can use AI to:

• test arguments

• explore counterpoints

• identify blind spots

• embrace uncertainty

• refine reasoning

His prompts look like:

• “Ask the AI to critique your explanation.”

• “What assumptions are you, or the model, making here?”

• “Find a counterexample that challenges your classification.”

Metis doesn’t replace thinking. He provokes it.

He’s a light reminder that AI is most powerful when used deliberately, reflectively, and dialogically.

Four Practical Scaffolds You Can Add to Any Course

These are the scaffolds that consistently work across disciplines, and they don’t require rewriting your entire curriculum.

1. Clear Permission Levels

Students use AI more responsibly when they know exactly what’s allowed.

A simple four-level model works:

• AI-0: No AI

• AI-1: AI for brainstorming or light editing only

• AI-2: AI allowed, but students must verify and document

• AI-3: AI-integrated tasks where prompting is part of the learning

Clarity reduces anxiety and prevents accidental misconduct.

2. Require “Process Evidence,” Not Just Outputs

If assessment only rewards polished products, AI will produce polished products.

But if you require:

• drafts

• screenshots

• prompt logs

• annotations

• reflect-and-revise notes

…students must show their thinking, not hide behind a model’s fluency.

This is one of the strongest safeguards for learning integrity.

Check out my AI Disclosure Proforma

3. Teach Students to Challenge AI

Most students assume AI outputs are “correct enough.”

But higher education requires more than “correct enough.” It requires critical judgement.

Teach students to ask:

• “Where might this be oversimplifying?”

• “What assumptions shape this explanation?”

• “Whose perspective is missing?”

• “What uncertainties or disagreements exist here?”

These questions build what I call meta-AI literacy, the ability to interrogate AI as a partner, not a source of truth.

4. Encourage Structured Dialogue, Not One-Shot Prompts

The biggest shift educators can make is teaching students to stay in conversation with AI.

A good structure is:

1. Attempt your own answer

2. Ask AI to critique it

3. Defend or refine

4. Seek counterexamples

5. Summarise how your understanding changed

This moves students from consumption to co-construction.

A Simple Example You Can Use Tomorrow

Task: Classify generative AI as sustaining, disruptive, or transformative.

Scaffolded version:

1. Student writes their classification first.

2. Student asks AI:
   “Where are the weaknesses in my explanation?”

3. Student challenges the model:
   “Provide a counterexample from a regulator’s perspective.”

4. Student revises their classification.

5. Student submits a short reflection describing what shifted in their understanding.

This takes a simple task and turns it into a thinking exercise.

Why All of This Matters

Our students are entering a world where generative AI will sit alongside them in every intellectual task they perform.

If they learn to rely on it uncritically, they risk weakening their judgement, their reasoning, and their intellectual independence.

But when we scaffold AI use intentionally, something powerful happens:

• students think more deeply

• their reasoning becomes more defensible

• their confidence grows

• and they learn to collaborate with AI rather than outsource their thinking to it

This is the future of responsible AI use in higher education.

Not avoidance.
Not overuse.
But deliberate, transparent, structured engagement.

Want More Guidance Like This?

I’m currently completing my book, The AI Educator, which explores scaffolding, assessment redesign, ethical AI literacy, and practical teaching strategies for the era of generative AI.

If you’d like to be notified when it’s released, subscribe here, I’d love to keep the conversation going.

Subscribe now

Because what truly deepens learning isn’t just making something, it’s talking about it: explaining, questioning, revising, and refining our ideas in dialogue with others.

This is where Diana Laurillard’s Conversational Framework (2012) becomes a compass for teaching in the age of AI. It helps us see education not as the transfer of information but as a cycle of communication between teacher and learner, learner and peer, and increasingly, learner and machine.

The Essence of the Conversational Framework

Laurillard proposed that effective learning follows a dynamic dialogue:

1. Conceptual Exchange: The teacher presents ideas; the learner interprets and responds.

2. Practice and Feedback: The learner applies knowledge; feedback refines understanding.

3. Reflection and Adaptation: Both teacher and learner adjust their models of thinking.

This loop repeats across levels of abstraction, connecting theory to experience.
In this model, teaching is never static it’s a continuous conversation about meaning.

When we embed AI into this cycle, the goal isn’t to replace either voice but to extend the conversation by giving learners more opportunities to articulate, test, and reconstruct understanding.

Example 1: AI as Reflective Partner

In a psychology course, students might discuss ethical dilemmas using an AI companion trained on professional guidelines.

Rather than offering “right” answers, the system poses counter-questions:

“What ethical principle might conflict with that choice?”
“How would this decision change if the participant were a minor?”

The AI doesn’t instruct; it provokes reflection.
Students refine their reasoning in response and teachers later review transcripts to identify moments of cognitive dissonance worth unpacking in class.

Here, AI acts as a conversation starter, helping students rehearse reasoning before bringing it into the social space of discussion.

Example 2: The Adaptive Feedback Loop

In a chemistry lab simulation, an AI coach monitors experiment logs. When a student adjusts variables inconsistently, the system intervenes:

“You changed temperature but not concentration. What outcome do you expect from that?”

The student must articulate a prediction before proceeding. After observing results, the AI prompts:

“Did the outcome match your expectation? Why or why not?”

This back-and-forth keeps the learner in a reflective stance, turning trial and error into deliberate experimentation; dialogue as inquiry.

Example 3: Scaling Feedback Without Losing the Human

One of the hardest challenges in higher education is maintaining rich feedback at scale.

Imagine a writing course where AI provides first-pass formative feedback framed in conversational terms:

“You’ve argued strongly for economic reform here. What counter-arguments might strengthen your conclusion?”

It creates talking points for tutorials. Teachers then step in to extend the conversation guiding tone, nuance, and judgment while saving hours on mechanical corrections.

Instead of replacing feedback, AI helps teachers spend their energy where it matters most: interpreting meaning and mentoring growth.

The Teacher’s New Role: Dialogue Designer

Within this framework, teachers become designers of conversation. They decide:

• What kinds of questions the AI should ask.

• Where to hand the conversation back to human discussion.

• How to assess reflection, not just response.

This design work demands the same intentionality as lesson planning aligning each AI interaction with learning outcomes and ethical values. It ensures that technology remains a facilitator of thinking together, not a dispenser of canned replies.

The Ethical Edge of Dialogue

Dialogue brings responsibility. When AI participates in conversation, it also shapes perspective. We must therefore attend to:

• Bias: Which voices are amplified or silenced by the model’s training data?

• Transparency: Do students know when they’re engaging with an algorithm?

• Privacy: Who owns the dialogue traces that record their thinking?

Responsible AI dialogue isn’t just about civility; it’s about agency ensuring learners remain authors of their own ideas. As Laurillard reminds us, genuine learning dialogue “requires that each participant has a voice that can influence the other” (2012, p. 86). The same must hold true when one voice is synthetic.

Wrapping Up: From Data to Dialogue

Across this series, we’ve moved from diagnosing the problem to imagining solutions:

• In Part 1, we saw that AI often misreads learning because it overlooks emotion, motivation, and metacognition.

• In Part 2, we embraced Generativism, using AI to foster creation and reflection rather than compliance.

• And now, in Part 3, we’ve grounded those ideas in conversation which are the heartbeat of authentic education.

The next frontier isn’t smarter algorithms. It’s smarter conversations; ones where teachers, learners, and AI systems co-construct understanding through curiosity, feedback, and care.

References

Laurillard, D. (2012). Teaching as a design science: Building pedagogical patterns for learning and technology.Routledge.

So where do we go from here?

If Part 1 was about what AI misunderstands, Part 2 is about what we can teach it by designing and using AI that encourages learners to generate knowledge, not just reproduce it. This is the heart of Generativism: the idea that learning happens through creation, explanation, and transformation, not passive consumption.

The Generative Mindset

Generativism is not a new concept. In fact, it traces back to Wittrock’s (1974) insight that learners actively build meaning by linking new ideas to prior knowledge. But its significance has sharpened in the age of generative AI.

As Mairéad Pratschke (2024) argues in Generative AI and Education, generativism has evolved into a digital pedagogy which presents a framework for designing, delivering, and assessing learning in collaboration with AI tools. In this view, generative learning is no longer just about students producing knowledge; it’s about co-creating understanding with machines, using AI to extend exploration and imagination.

When AI is treated as a partner in inquiry, not a producer of answers, learning becomes dialogic again. The process shifts from using AI to thinking with AI.

Example 1: Generative Writing Feedback

Traditional writing assistants focus on surface-level polish including grammar, clarity, tone. A generative approach looks deeper.

Imagine an AI feedback tool that doesn’t just correct a sentence but prompts the student to explain their reasoning:

“You used the phrase ‘this shows that’. What connection are you trying to make here?”

It could then offer scaffolded questions:

“How does this example strengthen your argument?”
“What counterpoint might challenge this idea?”

Each prompt draws the student back into meaning-making. Instead of being a shortcut to better prose, AI becomes a mirror for thought. The student writes, reflects, and revises. This is learning through articulation, not automation.

Example 2: Student-Generated Quizzes

In a generative classroom, learners don’t just answer questions, they write them.

Picture a platform where students use an AI assistant to create quiz questions for their peers. The system analyses each item, identifies whether it tests recall, inference, or evaluation, and suggests how to deepen it:

“This question focuses on factual recall. Could you reframe it to ask why or how?”

Students learn to think like assessors, anticipating misconceptions and clarifying concepts. This not only reinforces their own understanding but fosters metacognitive awareness, the realisation that how we question is as important as what we know (Chi & Wylie, 2014).

Example 3: Creative Co-Design with AI

In a design or media course, students might use AI tools to explore variations on a theme by generating images, scripts, or prototypes.

A generative approach doesn’t end with the AI output; it begins there. Students critique the results: Which version best reflects my intent? What assumptions did the AI make? How could I refine my prompt to express a clearer vision?

This recursive cycle of prompt → response → reflection → revision, turns creative production into a metacognitive exercise. It’s not just about making artefacts; it’s about understanding the relationship between ideas, tools, and judgment.

From Prediction to Participation

These examples share a common thread: AI becomes a participant in learning, not a predictor of outcomes.

Predictive AI asks, “What will the learner do next?”
Participatory AI asks, “How can we help the learner think next?”

Generativism (as presented by reframes AI as a conversational partner in which it prompts, challenges, and supports the learner’s own cognitive effort. The emphasis shifts from efficiency to exploration, from correctness to curiosity.

Why Generativism Matters Now

As schools rush to adopt generative AI, the temptation is to use it for convenience: faster marking, instant lesson plans, automatic summaries. Those efficiencies matter, but they don’t transform learning.

The real transformation happens when we use AI to open cognitive space, not close it. When it helps students explain, question, and make connections, it supports the very processes of motivation, emotion, and metacognition, that Part 1 showed AI has historically neglected.

That’s the promise of the Generative Turn: to make learning active again.

Steps for Educators

1. Recast AI tools as thinking partners.
   Ask how a tool can provoke reasoning, not just polish products.

2. Build generative tasks.
   Replace “write a summary” with “use AI to generate three interpretations and critique them.”

3. Teach prompt refinement as reflective practice.
   Show students that changing a prompt isn’t gaming the system. Instead it’s a form of inquiry.

4. Reward the process, not the product.
   Assess how learners use AI to explore ideas, not just the final output they submit.

Coming Next: The Conversational Classroom, Building Dialogue with AI

In the final part of this series, we’ll look at Laurillard’s Conversational Framework; a model for turning generative principles into structured dialogue. We’ll explore how AI can take part in that conversation: prompting reflection, supporting feedback loops, and helping teachers scale meaningful interactions without losing the human voice at the centre.

References

Chi, M. T., & Wylie, R. (2014). The ICAP framework: Linking cognitive engagement to active learning outcomes. Educational psychologist, 49(4), 219-243.

Hooshyar, D., Šír, G., Yang, Y., Kikas, E., Hämäläinen, R., Kärkkäinen, T., ... & Azevedo, R. (2025). Towards responsible AI for education: Hybrid human-AI to confront the Elephant in the room. arXiv preprint arXiv:2504.16148.

Laurillard, D. (2013). Teaching as a design science: Building pedagogical patterns for learning and technology. Routledge.

Pratschke (2024). “Generativism.” In Generative AI and Education: Digital Pedagogies, Teaching Innovation and Learning Design (SpringerBriefs in Education), Chapter 4, pp. 57–72.

Wittrock, M. C. (1974). Learning as a generative process. Educational psychologist, 11(2), 87-95.

• Part 1: When AI Gets Learning Wrong

• Part 2: The Generative Turn — From Prediction to Participation

• Part 3: The Conversational Classroom — Building Dialogue with AI

Part 1

Artificial intelligence has become education’s favorite new assistant. It grades essays, generates quizzes, and tracks progress with a precision no spreadsheet could match. Yet beneath the surface, a question keeps nagging at me:

Does AI actually understand learning—or just measure it?

A recent publication by Hooshyar and colleagues (2025) suggests the latter. The authors argue that most AI-driven learner-modeling systems “ignore essential learning processes such as motivation, emotion, and (meta)cognition and their contextual nature.” In other words, AI can see behaviour but not learning.

That insight explains why so many “intelligent” systems still misread what’s really happening in classrooms.

The Missing Middle: What AI Can’t See

Take Amira, a student who completes every algebra exercise on her laptop. The dashboard lights up with green check marks, signaling mastery. What it doesn’t show is the quiet anxiety that keeps her glued to procedural routines where she’s memorising, not understanding. The AI’s analytics celebrate her accuracy, but they can’t sense the hesitation behind each click or the growing fear of making a mistake. To the system, she’s thriving. To her teacher, she’s surviving.

Then there’s Luca, whose screen time looks like disengagement. He hesitates, stares into space, doodles in his notes and then types a single, brilliant explanation that connects two concepts no worksheet ever asked him to compare. His learning is slower, reflective, meandering. It’s the kind of thinking that needs space and silence.

For the algorithm, Amira is a success and Luca a problem. The data paints her as efficient and him as distracted. But for a teacher, the story is far more complex: one student is achieving without understanding, the other is struggling toward insight. What looks like productivity on a dashboard may in fact be a symptom of fear; what looks like idleness may be the visible form of deep thought.

AI can’t see any of that because its models treat learning as a linear path to mastery, not the messy, emotional, self-correcting journey that it really is.

The Cognitive Bias of Our Machines

This isn’t just a technical oversight; it’s a philosophical one. Learning is not a linear data stream but a recursive process shaped by feedback, emotion, and context (D’Mello & Graesser, 2023).

Yet AI systems are typically built on static correlations. They assume that if “practice leads to the correct answer,” learning has occurred. But research in educational psychology shows that mastery is inseparable from metacognition which is the learner’s awareness and regulation of their own thinking.

When AI ignores that dimension, it risks rewarding surface performance and penalizing productive struggle. Students who learn reflectively may look inefficient to a machine trained on speed and accuracy.

A Richer Model of Learning

To fix this, we need AI that reflects the complex ecology of learning and its cognitive, emotional, motivational, and social layers.

Imagine an intelligent tutor that doesn’t simply mark an answer wrong but asks,

“You changed strategies here. What made you decide to try that?”

The system might detect hesitation or frustration in a student’s text responses and prompt encouragement or self-explanation instead of another hint. That’s not science fiction; it’s a design philosophy grounded in responsible, hybrid human–AI approaches (Hooshyar et al., 2025).

Hybrid systems blend data-driven modelling with expert knowledge from educators and psychologists, ensuring that algorithms are guided by real learning theory, not just pattern recognition.

What Educators Can Do Now

While we wait for better tools, teachers can begin reclaiming the conversation about AI and learning by asking three simple questions about any technology they use:

1. What kind of learning does this system see?

   If it only measures performance, pair it with human reflection tasks that capture emotion and reasoning.

2. How does it handle motivation?

   Does it encourage curiosity and persistence, or merely completion?

3. Does it support metacognition?

   Are students prompted to explain their thinking, set goals, or evaluate their own progress?

When we start asking these questions, we move from data collection to dialogue creation.

Why It Matters

Education is not a data-processing problem; it’s a meaning-making process. As Hooshyar and colleagues warn, AI built without a human model of learning can reinforce inequities and misclassify students who don’t fit statistical norms.

Our task as educators is not to resist AI, but to teach it what learning truly looks like which is messy, emotional, and profoundly human.

Reflection Prompt for Educators

Where in my teaching do I already capture learners’ motivation or metacognition and how could AI tools help me amplify those insights instead of replacing them?

Next up: The Generative Turn, From Prediction to Participation

In Part 2, we’ll explore how educators can design AI not as a grading assistant but as a creative collaborator. We’ll unpack the principles of Generativism or how learning deepens through creation, and see how teachers are using AI to help students generate, not just consume, knowledge.

Because if Part 1 asks what AI gets wrong about learning, Part 2 will ask how we can teach it to get learning right.

References

D’mello, S., & Graesser, A. (2013). AutoTutor and affective AutoTutor: Learning by talking with cognitively and emotionally intelligent computers that talk back. ACM Transactions on Interactive Intelligent Systems (TiiS), 2(4), 1-39.

Hooshyar, D., Šír, G., Yang, Y., Kikas, E., Hämäläinen, R., Kärkkäinen, T., Gašević, D., & Azevedo, R. (2025). Towards responsible AI for education: Hybrid human-AI to confront the elephant in the room. arXiv preprint arXiv:2504.16148.

In Chapter 6 of The AI Educator, I write about how tools like ChatGPT risk turning “an opportunity to wrestle with complex ideas into a button-click exercise in outsourcing thought.” That line has resonated with many educators because it speaks to something deeper than academic integrity. It’s about what it means to learn.

We often tell students, “Do the hard work.” But what’s really happening in the brain when learning feels hard? And why does that discomfort matter?

The Neuroscience of Struggle

When learners tackle difficult material, the prefrontal cortex, the brain’s control centre for working memory and planning, activates intensely. This region keeps information “in play,” suppresses distractions, and coordinates reasoning (Miller & Cohen, 2001).

Recent imaging studies show that when people expect a task to be difficult, the prefrontal cortex sustains activation longer, indicating greater mental effort (Master. S et. al., 2024).

In short, the brain quite literally works harder when it anticipates challenge.

This kind of “cognitive strain” isn’t wasted energy; it’s when the brain is forming new neural connections. Cognitive load theorists describe learning as the management of mental effort within the limits of working memory. When tasks are slightly beyond our comfort zone, the brain integrates new information with prior knowledge. This is what psychologists call germane load and it strengthens long-term understanding (Paas & Sweller, 2014).

So when students struggle, their brains are reorganising networks and encoding deeper representations. The feeling of effort is actually evidence of growth.

Why Effort Feels Uncomfortable

Neuroscientists have also found that the brain treats cognitive effort as a kind of cost. The anterior cingulate cortex weighs how much effort a task will take against its perceived reward (Kool & Botvinick, 2018). That’s why difficult work feels aversive, not because students are lazy, but because their brains are managing limited resources.

Yet that very resistance marks the edge of development. Pushing through that “ugh” moment engages dopaminergic reward circuits once progress is made. Over time, the brain learns to associate effort with accomplishment. In other words, struggle rewires motivation itself.

Designing for the Hard Work

What does this mean for classrooms navigating the rise of AI? If AI smooths away all friction, we risk losing the conditions that make learning transformative.

Here are a few strategies to keep the “good struggle” alive:

• Make thinking visible. Ask students to submit drafts, prompt histories, or reflection notes explaining their decisions. Assessment becomes a record of thought, not just output.

• Reframe difficulty as progress. Remind learners that the “stuck” feeling signals new neural growth, not failure.

• Balance challenge and support. Tasks that are too easy don’t engage the prefrontal cortex; too hard, and motivation collapses. Scaffold so students operate in the “productive difficulty” zone.

• Reward reasoning, not polish. Grade the clarity of decision-making, evidence of revision, and depth of reflection.

When students understand that effort is where learning happens, frustration transforms into persistence.

Try this:
Next time you teach with AI, ask students to include a short reflection:

“What part of this task felt hard and how did you push through it?”

That’s where the most powerful learning insight usually hides.

AI can generate, summarise, and polish, but it can’t do the truly human work of deciding what matters. When students outsource all the heavy lifting, they miss the very process that builds intelligence: sustained attention, reflection, and conceptual struggle.

As I argue in The AI Educator, the goal isn’t to reject AI but to design learning experiences where AI assists without erasing the thinking journey. The hard work is the learning and it’s that part that makes us more human.

Subscribe now

References

Kool, W., & Botvinick, M. (2018). Mental labour. Nature human behaviour, 2(12), 899-908.

Master, S. L., Li, S., & Curtis, C. E. (2024). Trying harder: how cognitive effort sculpts neural representations during working memory. Journal of Neuroscience, 44(28).

Miller, E. K., & Cohen, J. D. (2001). An integrative theory of prefrontal cortex function. Annual review of neuroscience, 24(1), 167-202.

Paas, F., & Sweller, J. (2014). Implications of cognitive load theory for multimedia learning. In The Cambridge handbook of multimedia learning (pp. 27-42). Cambridge University Press.

What the Data Says

TALIS 2024 captured the perspectives of roughly 300,000 teachers and their principals across more than 50 education systems worldwide. Each participating country sampled around 4,000 teachers from 200 schools per education level, covering primary (ISCED 1), lower secondary (ISCED 2), and upper secondary (ISCED 3) teachers.

This means TALIS provides a remarkably comprehensive snapshot spanning classroom teachers from early years through upper secondary education and offers one of the clearest pictures yet of how educators are encountering AI in practice.

TALIS 2024 shows that around one in three teachers across OECD countries have used AI tools in their teaching. The numbers vary dramatically: 75% of teachers in Singapore and the UAE report using AI, compared to fewer than 20% in France and Japan.

Among those who do use AI:

• 68% use it to learn about or summarise topics,

• 64% use it to generate lesson plans,

• 25% use it to analyse student data or participation, and

• 26% use it for grading or assessment.

Teachers see both promise and peril. About 40% agree that AI helps them support students individually, while seven in ten worry that AI enables plagiarism or cheating.
Perhaps most concerning: three in four teachers say they lack the skills or knowledge to teach effectively with AI, and half don’t believe AI should be used in teaching at all.

So, while headlines might suggest an AI revolution sweeping through classrooms, the reality is much more cautious with a blend of curiosity, experimentation, and apprehension.

The View from The AI Educator

In The AI Educator, I take a slightly different lens. I’ve never seen AI adoption as a binary of “use it or don’t.” Instead, I argue that what matters most is how we use it, and more importantly, the values guiding that use.

In Chapter 2, for instance, I describe teachers as knowledge designers, not content deliverers. The book frames AI as a thinking partner, as a collaborator that can amplify human judgment, not automate it.

The OECD findings highlight a troubling imbalance: AI is being used most often for efficiency tasks meaning it’s used for summarising content or generating lesson plans rather than pedagogical innovation. This suggests many teachers are still using AI to save time, not necessarily to transform learning.

That’s understandable. But if we stop there, we miss AI’s deeper potential: to help students become critical co-creators of knowledge, not passive consumers of machine output. As I write in Chapter 8, “AI is neither a miracle solution nor a looming threat, but a tool whose value depends on how thoughtfully it is woven into the fabric of education.”

Where We Go from Here

The TALIS data tells us teachers are cautiously stepping in but often without enough guidance or training. In contrast, The AI Educator calls for teachers to step up and lean in and take ownership of AI’s integration through ethical reflection, creative experimentation, and community dialogue.

We can’t just focus on using AI to make teaching easier; we need to focus on how it can make learning richer. That requires more than digital literacy. It requires pedagogical imagination.

As we move forward, the challenge is not simply to increase the number of teachers using AI, but to support them in using it with purpose including a grounding in trust, empathy, and human-centred design. Because in the end, as I argue throughout the book, AI may change how we teach, but it will never change why we teach.

If you’ve ever coordinated assessment across a large class, you know the struggle. Multiple tutors, each with their own experience level and marking style. Time pressures that make it hard to offer quality feedback. Students frustrated by comments that feel generic, or worse, unfair. It’s a perfect storm that leaves everyone feeling dissatisfied.

Marking at scale isn’t just about consistency; it’s about care. Students deserve meaningful, individualized feedback, and tutors deserve systems that make this possible without burning them out. This is where large language models (LLMs) can quietly transform the marking process; not by replacing human judgment, but by amplifying it.

The first part of this post is about a system I created myself, but as you will find out at the end, the same thing can be achieved with ChatGPT.

Aligning AI Feedback with Rubrics

The idea is simple: instead of leaving feedback entirely to either humans or machines, we can align both.

Tutors select the relevant rubric criteria, either through checkboxes or numerical scores, and an LLM translates those selections into personalised, context-aware comments. Optional student work, such as code or essays, can also be incorporated for finer feedback. The result is feedback that’s consistent, individualized, and grounded in the same rubric that underpins assessment.

This approach keeps tutors in the loop but removes the friction of repetitive comment writing. It also ensures students receive feedback that feels authentic and fair. Something grounded in both human evaluation and linguistic nuance.

Building the System: My Prototypes

Over the past year, I experimented with this concept by developing simple webpages hosted on my own server, all connected to the OpenAI API. The goal was to see how far AI could go in improving marking workflows.

Prototype 1: Fully Automated (GPT-3)
The system provided the AI with the rubric, assignment description, and student files. GPT-3 generated both marks and feedback automatically. While quick, it was far too generous and occasionally wrong. The tone was polished, but the accuracy wasn’t there.

The rubric and assignment description files were stored on the web server. All the markers had to do was drag and drop the student submission onto the webpage.

After pressing submit and a little waiting time, the AI would respond with an evaluation of the assignment as shown here:

Prototype 2: Iterative Automation (GPT-4)
This version improved by first asking the model to apply the rubric, then generate marks and feedback in a two step process. The results were better in that they were more balanced and more personal, but still inconsistent in edge cases.

Prototype 3: Human in the Loop
The breakthrough came when I combined human judgment with AI assistance. Human markers completed the rubric first, and the model used that input to generate contextualized feedback. The comments aligned tightly with both the scores and the assignment description. The difference was immediate: accuracy up, marking fatigue down, and tutors reported feeling more confident in their feedback.

The results after submission:

Using ChatGPT instead (or any other LLM)

For those of you without the coding now how, I was interested in determining if the same process could be done using ChatGPT. And it can! Just by submitted a spreadsheet with the completed marking.

The spreadsheet, that you can get via this link includes the prompts, rubrics and human assessment marks. This is simply copy and pasted into ChatGPT and what results is a nice little report, which the markers can then pick and choose the feedback from to provide students with the greatest personalised and relevant feedback.

See how it can be done in this video:

What We Gain

1. Tutors and Markers

• Reduced fatigue through automation of repetitive phrasing

• Greater consistency across multiple markers

• Focus on judgment, not wording

As Willey and Gardner (2010) put it:

“Engaging in structured marking practices enhances tutors’ confidence, fosters professional growth, and promotes reflective teaching.”

2. Students

• Timely, specific, and consistent feedback

• A clearer understanding of performance criteria

• Improved motivation and self-regulation

Foster (2024) found that:

“Formative assessment practices that deliver timely, specific, and consistent feedback significantly improve student learning outcomes, motivation, and self-regulation.”

Impacts and Insights

In practice, the system produced:

• Increased tutor confidence and alignment

• Faster turnaround times for returning results

• A scalable framework for consistent, personalized feedback in future course offerings

And perhaps most importantly, it reframed AI not as a shortcut, but as a support structure. One that helps humans focus on what they do best: judgment, mentoring, and care.

Closing Reflection

As I’ve written in The AI Educator, automation in education should never be about replacing teachers. It’s about reclaiming time and clarity. When used thoughtfully, AI can take over the heavy lifting of routine tasks so educators can spend more time engaging with students, not spreadsheets. The goal is not to make marking faster, but to make feedback fairer, more human, and more meaningful at scale.

Because in the end, it’s not just about the marks as much as it’s about learning.

References

Foster, H. (2024). The impact of formative assessment on student learning outcomes: A meta-analytical review. Academy of Educational Research Journal, 28(1), 15–32.

Willey, K., & Gardner, A. (2010). Improving the standard and consistency of multi-tutor grading in large classes.University of Technology Sydney.

Most institutional frameworks now include AI disclosure requirements. Students are expected to declare which tools they used, show how outputs were checked, and reflect on the learning impact. Yet, in practice, many students (and teachers) are left wondering what that actually looks like.

That’s why in Chapter 5 of The AI Educator (in press) I’ve included a simple resource to help: the AI Use Disclosure Pro Forma. It’s a one-page document students can attach to any assessment where AI tools played a role.

Rather than adding bureaucracy, it aims to make learning visible. Students list tools, describe how they interacted with AI, and reflect on what they learned in the process. The structure encourages transparency without fear by helping students own their learning choices and demonstrate integrity in a world where AI is everywhere.

At its core, this form guides students through three key questions:

1. What did I use, and why?

2. How did I verify or adapt what the AI produced?

3. What did I learn about the process (including its limits or biases)?

These are not compliance questions; they’re reflection prompts. They help students articulate how AI supported their thinking and where their own judgment came in.

When we talk about ethical AI use, this is what it looks like in action: transparent, traceable, and teachable.

You can download the AI Use Disclosure Pro Forma here and use it in your courses. Encourage students to complete it alongside drafts, design logs, or reflections. It works equally well for essays, projects, and creative work.

Because rules tell students what to disclose.
Reflection teaches them why it matters.

If you use the form as is or altered please let me know!!

Well, for starters it’s a departure from the norm.

After years of writing sci-fi novels and textbooks on game development, I’ve taken a new turn. This book isn’t fiction, and it’s not a how-to manual either. It’s about something that’s reshaping the domain I’ve lived and worked in for over three decades: how artificial intelligence is transforming education and what we, as educators, should do about it.

Not a Crisis, But a Catalyst

Now, that might sound like the setup for an advocating a flat out ban. But quite the opposite is true. In all my years as a teacher and academic, I’ve never seen a disruption quite like generative AI. The speed and scale of its impact have been extraordinary and so has the scrambling that’s followed: banning it, policing it, experimenting with it, and, for some, learning to embrace it.

And let’s be honest, the biggest focus has been on cheating.

Every day, there’s a new headline about students using AI to complete their assignments. Of course, students have always found ways to cut corners, but GenAI has made that temptation frictionless. More importantly, it’s exposed something deeper: that our traditional ways of assessing learning simply don’t hold up anymore. Essays, reports, and un-invigilated tests no longer guarantee evidence of understanding or critical thought.

These concerns reveal something important: educators aren’t just afraid of AI itself. We’re uneasy about what it reflects: the cracks in our systems and the outdated assumptions about how learning works.

But that’s a conversation for another post.

Motivation #1: Understanding the Disruption

My first motivation for writing this book was to truly understand what we’re dealing with. I’ve worked with AI for much of my career and even created a course on Udemy about how to use generative AI. But its impact on pedagogy and curriculum was still a blurry area and I know many educators feel the same way.

Through the research and writing process, I learned a lot. More importantly, I discovered a growing community of researchers and teachers who are doing remarkable work not just studying AI’s disruption, but experimenting with how to integrate it meaningfully into their practice. They’re not running from it; they’re leaning into it.

Motivation #2: Empowering Educators

That brings me to my second motivation: to share that research and practice with you.

There’s been no shortage of calls for educators to embrace AI, but very little concrete advice about how to actually do that. What should it look like in the classroom? How do we use it without losing our pedagogical compass?

My goal with The AI Educator is to help answer those questions. To empower teachers with the understanding, tools, and confidence to use GenAI with purpose and integrity. It’s not about replacing teachers; it’s about reimagining teaching in a world where AI is part of the learning process.

Beyond the Essay Prompt

Sure, you can ask ChatGPT to write a decent essay on the Industrial Revolution and get something serviceable in return. But that’s not where its power lies. The real strength of generative AI is in the dialogue; the back-and-forth that challenges your thinking, refines your ideas, and helps you see concepts from new angles.

That’s where the real learning happens.

Looking Ahead

Over the coming weeks, I’ll be unpacking key ideas from The AI Educator. From the myths and fears surrounding AI in schools, to the redesign of assessment, to practical ways teachers can bring AI into their classrooms with confidence and care.

Because this isn’t about AI replacing us.
It’s about us redefining what great teaching looks like in the age of AI.

Follow along for more reflections, resources, and practical insights from The AI Educator: A guide to teaching, learning, and thriving in the era of generative AI.