Welcome! Before we get into the next in the “Artificial Intelligence, Real Literacy” series, I wanted to invite you to my upcoming PD.

We’re teaching and learning in the midst of the most rapid change the world has ever seen. But what do our teachers actually need to use to be capable users of AI? Without AI capability we fail to explore the many good possibilities while falling victim to the many problems AI has introduced.

But teacher AI capability is not rocket science. In fact, in this PD, I’ll show you a simple, four-step framework. It’s clear, actionable, and I think you’ll love it.

There’s a link below - I hope to see you there!

The writing technique that changed everything for me

I love writing.

Over the last three years, it’s changed my life. Weekly long-form posts, daily content on LinkedIn. I’ve written around 550,000 words. That’s roughly six Harry Potter books.

And what I’ve appreciated about this daily writing habit is not the formation of text, it’s the formation of myself. Years of writing about AI, education, and pedagogy, have left me with a clear philosophy and actionable practices that benefit myself and others.

But here’s the thing.

At school I was a terrible writer. Actually terrible.

I could think. I could talk. But getting ideas on the page was harder than the gingerbread man I found under my 4-year-old’s bed last week. The words would come out mangled or I’d stare at a blank page and feel stuck.

The technique I’m about to show you played a big part in changing that.

It’s called sentence stems, and it’s one of the most powerful writing scaffolds I know.

What is a sentence stem?

A sentence stem is a partially completed sentence designed to help a struggling learner get their thoughts onto paper. We provide the stem, and the student branches out with their own thinking.

There are two kinds of stems that I use:

Single-blank stems give you one gap to fill:

• I thought the main theme of the story was _____.

• To solve this equation, first I have to _____.

• This discovery could impact our lives by _____.

Multi-blank stems ask you to make multiple connections:

• The author’s use of _____ symbolises _____.

• During the experiment, we observed that _____ caused _____.

• The _____ movement was a response to _____, leading to _____.

The logic is simple. When a struggling writer is learning something new, they’re fighting on two fronts at once: the new idea and the challenge of expressing it. A sentence stem removes one of those battles. It provides the structure so the student can focus on the thinking.

That’s exactly what it did for me. I always had the thinking. The stem gave me a place to start.

Why they work

There’s a concept in education called the Zone of Proximal Development. The idea is that good scaffolding lands right in the middle: not so much support that there’s nothing left to do, and not so little that the student is lost.

A well-written sentence stem puts the learner in that zone. It gives them just enough to begin.

Keely Keller from the Teaching Channel puts it well: sentence stems shift the focus from the mechanics of writing to the content you actually want students to engage with.

And for a struggling writer, that’s going to make all the difference.

How to use AI to create them

Using AI to create sentence stems is a great application of the vision I cast at the beginning of this series: using AI to do the basics better. So here’s the prompts that I use to create sentence stems with AI tools like ChatGPT, Copilot, or Claude.

Generic stems (great for lesson introductions or retrieval practice at the end of class):

Role: Act as an expert educator and curriculum designer.

Task: Create a list of sentence stems on the topic of [INSERT YOUR TOPIC]. A sentence stem is a prompt or opening phrase that helps students begin their sentences or ideas, facilitating writing and discussion by providing a structured starting point. Include a mix of single-blank stems (one gap to fill) and multi-blank stems (two or three gaps), and vary the cognitive demand across the list.

Format: Give me 10 sentence stems in bullet points.

Specific stems (tied to a particular text, video, or image):

Role: Act as an expert educator and curriculum designer.

Task: Create a list of sentence stems based on the text I’ve attached. Limit the stems to ideas and language from this source only. A sentence stem is a prompt or opening phrase that helps students begin their sentences or ideas, facilitating writing and discussion by providing a structured starting point. Include both single-blank and multi-blank stems.

Format: Give me 10 sentence stems in bullet points.

Differentiated stems (for classrooms with a wide range of learners):

Role: Act as an expert educator and curriculum designer.

Task: Create a set of sentence stems on the topic of [INSERT YOUR TOPIC]. A sentence stem is a prompt or opening phrase that helps students begin their sentences or ideas, facilitating writing and discussion by providing a structured starting point.

Format: Give me three stems under the heading “Simple,” three under “Moderate,” and three under “Complex.” For Complex stems, add a note that students may need more than one sentence to complete the idea. Include a mix of single-blank and multi-blank stems throughout.

You can also ask for stems organised by Bloom’s taxonomy if you want to build a full progression from recall through to evaluation.

A few things worth thinking about

If your students haven’t used sentence stems before, use the gradual release model. Model it yourself first. Then do one together. Then small groups. Then solo. Don’t just hand them a list and hope for the best.

For your more advanced students, flip it. Ask them to write the stems. That’s a genuinely challenging metacognitive task and a great extension activity.

And at the end of a writing period, get students to share their completed stems with each other. The discussion that follows is often where the real consolidation happens.

Sentence stems are also one of the quickest and most reliable ways to do informal formative assessment. A handful of completed stems tells you a lot about how deeply a student has understood something.

So, that’s a quick guide to using AI to create sentence stems.

We’ve always known that they benefit our learners, but now the time cost taken with creating high quality resources has been largely removed. We’re living in an age where we can do the evidence-backed basics more often for more learners, and I’m not sad about it!

So why not give it a go this week?

Use AI to create a simple resource that could help your struggling writers make a start.

Let me know how you go!

Hey friend,

I trust you’ve had a great week so far.

I had a great surprise last weekend when I saw that Wise AI got featured in my local paper. This combined with a bonfire, family movie night, and seeing my father give a presentation meant it was a great finish to the week!

Before we get into the essay this week, I want to tell you about something really exciting.

I’m running professional development on my AI Capability for Teachers framework.

It’s designed to be clear, practical, and useful to school leaders and teachers. If you’d like to sign up, just click the banner below.

(And if you can’t make the time slot - I’ll be making the recording available to everyone who registers!)

Hope to see you there :)

Introduction

I feel sorry for our students.

I teach in a high school, and as I see my students walk around the campus, there’s a part of my heart that aches for them. And here’s what stirs my compassion:

They are students in the midst of the most rapid change in all of human history.

AI is a huge disruptor. There has never been change of this speed and scale.

I remember being in high school - it was a tricky enough time to navigate without the low-level anxiety that you’ll graduate into a world where a machine can do every job to a higher standard and for less money all while you sit on the couch eating potato chips collecting your universal basic income.

And make no mistake about it - our learners are wondering about these things. Some are hopeful, some are fearful, and many are taking this uncertainty about the future and using it as an excuse to kick the clutch in with their studies.

I understand how this change and uncertainty could manifest in educational lassitude. I do. But today I want to set the record straight about one big lie I keep hearing.

The big lie I keep hearing from students

Would you be alarmed to hear there’s a gangrenous lie eating away at our students?

They catch philosophical infection from their feeds or their friends, but I’ve seen symptoms of it everywhere. Here’s the lie:

We don’t have to learn the things AI can do for us.

I’ve seen the lie in the wild hundreds of times. (Usually as I’m pushing the students to improve their academic skills.)

• “We shouldn’t learn to write essays; we can get AI to do it!”

• “We shouldn’t learn to research; we can get AI to do it!”

• “We shouldn’t learn to summarise; we can get AI to do it!”

So if this is the gangrenous lie, these three responses are part of the remedy.

1) No, you actually can’t get AI to do it.

This is the simplest and most basic pushback.

Let’s take the essay for example. To get AI to write a good essay you need

• Domain-specific knowledge

• Knowledge of language conventions

• A gut feel for what good writing looks like

Students don’t have that yet.

Well, maybe some do, but most don’t. And why would they? They haven’t put in the hard yards of thinking, learning, and skill acquisition yet.

But those academic hard yards are what will enable them to use AI well in the future.

Good automation rests upon deep understanding.

You simply can’t automate what you don’t understand.

So no, you can’t get AI to write good essays yet. Maybe one day, but you’ve got a lot of learning to do first.

2) Writing an essay isn’t about writing an essay.

The purpose of the essay isn’t to create an essay.

It’s to create a certain type of person.

A person who can

• Think clearly

• Argue their case

• Debate respectfully

• Weigh ideas and evidence

• Sit with a problem for a long time

None of this happens if you outsource the task to AI.

Even if you could create the work required, you’re not becoming the person desired.

Education isn’t just about information; it’s about student formation – and essays remain a great tool for forming learners.

3) It’s based on a faulty understanding of technology

This argument presupposes we shouldn’t do things that technology can also do. Many students hold this as an article of faith, yet when pushed, struggle to explain the logic.

And to be honest, I don’t think the logic is there. Let’s think about other areas of life:

• Do we stop jogging now we can drive?

• Do we stop lifting now we have forklifts?

• Do we stop painting now we have Photoshop?

The default assumption that we should stop doing things that AI can also do is how we achieve the worst possible outcome:

It’s how big tech creates small humans.

That’s a future I’m actively fighting against.

Conclusion

So, no, I don’t think we should stop teaching essay writing or requiring students to do essays.

Because they aren’t just about producing five paragraphs with an introduction and a conclusion. They are about producing the kind of person who can think, reason, and argue their case. They are about producing people who can wonder, ponder, and analyse. And not only are these skills important, they’ve never been more important.

That said, I know not everyone is a fan of the essay. What about you – do you see a future for the humble essay? Or am I simply clinging to a bygone era?

Let me know!

In this essay, I’ll go over some of the big blunders we can make with AI. If using AI wisely is something you’re into - and I think it is! - why not sign up to our Wise AI masterclass. It’s absolutely free and I think you’ll love it.

Also, this week I had the opportunity to speak at the Australian Christian Schools Library Network national conference. It was an absolute blast - working in a library sounds great! Few pics below :)

Subscribe now

Introduction

In the 20th century, French philosopher Paul Virilio uttered a line for our time:

When you invent the ship, you also invent the shipwreck; when you invent the plane you also invent the plane crash... Every technology carries its own negativity, which is invented at the same time as technical progress.

While I don’t often find myself quoting French philosophers, I do happen to think he is on to something.

By inventing AI, we have unintentionally birthed a shadow realm of AI-related pathology, such as AI-slop, AI-hype, AI-creep, AI-fatigue, AI-anxiety, AI-paralysis, AI-psychosis, AI-dependence, and more.

This isn’t an AI problem. It is a technology problem.

So, let’s think about how to avoid 5 of the huge blunders we can make with AI.

Blunder 1: Being a Luddite or a Lemming

AI has a funny way of polarising us.

Some are convinced every AI prompt draws the apocalypse another day closer, while others are champing at the bit to get the Neuralink (and might even get two in case one is a little slow). In Wise AI, we call these characters the Luddite and the Lemming. Not only do I see them on my various feeds, I see it at conferences and within staffrooms.

The way forward is actually to recognise that both the Luddite and the Lemming have something to teach us.

From the Luddite we learn caution and a recognition that every tool has a cost.

From the Lemming we learn agility, optimism, and openness.

And we navigate AI wisely when we hold the tension between the two, exercising both careful caution and hopeful agility.

Blunder 2: product first, not purpose first

Imagine you’re on a boat.

As you unhitch from the dock you decide to let the rudder flap away in the water and let the boat decide where it wants to go. Ludicrous, right? Yet when it comes to AI, so often we let our tools shape our priorities instead of being led by our purpose.

AI tools push us to go faster. The metaphorical rudder is locked in a direction that will push us towards working faster on more things for longer periods of time. This is what happens when we let the product lead.

Instead, we can be led by our purpose. That’s when we’ll use the technology wisely.

AI works best when we grab the rudder and steer ourselves away from the whirlpool of busywork and towards deep work, deep thought, deep rest, or deep relationships.

If we are not led by our purpose, we will be led by our products. And that’s not what anyone needs.

Blunder 3: laminated values

We’ve all walked through the mind-numbing experience of an organisational update of the vision and values, only to know that the new values will simply be laminated, put in the staff room, and remain completely inert.

Yet we can so easily do the same thing.

We articulate our purpose only to do the individual equivalent of laminating it, putting it next to the microwave, and letting it accumulate splatters of spaghetti sauce.

To use AI wisely our values have to be lived, not just laminated.

If we are wanting to guard space for deep work, we must be wary of immediately deploying time saved on busywork into more busywork.

Blunder 4: treating exposure as expertise

In Australia we have a sport called Cricket (to all my English friends - how good were the Ashes…right?)

Bowling a ball - the equivalent of pitching in baseball - is a complex, full body movement.

Learning to bowl was when I realised that practice doesn’t make perfect. Because, as it turns out, a bad bowling action doesn’t magically fix if repeated over the course of five years.

We might be exposed to AI. We might even use it all the time. But without an understanding of the principles and practices of wise AI use, we won’t become experts.

Some of these ideas include:

• How to prompt an AI tool

• How to personalise tools to suit you

• How (and when) to iterate and tweak

• Where in the workflow AI should be used

• How to use AI in a way that doesn’t diminish organic intelligence.

Just like practicing my bad bowling action didn’t make me a better bowler, simply using AI won’t increase our expertise. Skill matters.

Blunder 5: getting stuck in the technological mud

I’ve said it, you’ve said it, but no one said it better than Bob Dylan:

The times, they are a-changin.

AI is changing quickly.

Only a few days ago I was hearing about how AI was ‘great for writing but couldn’t do maths’. Meanwhile AI is getting right out on the skinny branches of human knowledge and in some cases even doing new maths.

Long story short, what was once true may be true no longer. Not because it was false, but because the world has changed. If we get stuck in the technological mud and fail to see the change around us, we don’t put ourselves in the best place to explore the possibilities and avoid the problems of AI.

We can avoid this blunder by adopting the same position we expect our students to adopt every day, that of a learner.

Conclusion

None of these blunders are really about the technology.

They are about us. They are about the posture we choose to bring to our technology. The Luddite and the Lemming, the product-led drift, the laminated values, the mistaking of exposure for expertise, and the feet stuck in the mud all share one root. They are what happens when we don’t use AI wisely.

My encouragement to you is to stay curious, stay humble, hold your values close and your tools loosely, and you will be well placed to enjoy what AI offers while sidestepping the worst of what it threatens.

May you use AI wisely!

• My bad memory!

• An overview of retrieval practice as a teaching strategy

• How to use AI to create retrieval practice quizzes (including an AI prompt)

• Other things to consider as you differentiate texts for your learners

• Two professional learning opportunities (one of them is absolutely free)

I struggled as a student.

In fact, if you’ve been reading my essays for any length of time, you’ll know my problem was probably that I didn’t struggle enough. My high school education was a double degree in off-topic conversation and staring out the window.

And one of the reasons for this is that I was convinced I had a poor memory. I had poor pedagogical self-esteem.

And while I completed my education over a century after Ebbinghaus posited his Forgetting Curve, review and retrieval wasn’t a big part of my study.

I always felt like I was groping around the edges of previously learned ideas, never quite able to lay hold of them with clarity.

A decade and a half later I found myself teaching students who were also convinced they had a bad memory. It was like staring in a mirror (except the reflection spoke in gen alpha creole - lock in!!). And that’s when I started to make retrieval practice quizzes a mainstay in my pedagogical toolkit. These simple quizzes weren’t just a measure of learning, they were a learning event in their own right, and I saw the results first hand. Students actually remembered things.

The only problem was they took so jolly long to write.

Twenty minutes to draft a 5 minute quiz seemed like a bad return on investment, and creating these resources often got lost in the noise of administrivia.

But here’s the good news - both for you and your students - it’s never been easier to create tailored, specific quizzes that help your learners remember and expand upon their knowledge. What took 20 minutes now takes less than 2 minutes.

And here’s a practical guide for how you can use AI to create retrieval practice quizzes.

What Is Retrieval Practice?

Simply put, retrieval practice is the act of pulling information out of your memory rather than pushing it back in. (I talked about this in my TEDx talk.)

This could look like a student trying to remember the causes of World War One, answer a cold-call across the classroom, or (most often in my case), a low-stakes quiz.

The most accessible version, and the one I keep coming back to, is the humble quiz. Two multiple choice questions and two short answer questions can be enough. It’s not about getting a formal grade for a report card. It’s giving the brain a chance to do the one thing that actually builds memory: go looking for something and haul it back up.

Here’s the part that took me years to understand.

The quiz isn’t the test of the learning. The quiz is the learning.

When a student retrieves a fact, they’re not just proving they know it. They’re physically strengthening the path back to it, making it easier to find next time. The struggle to remember is the workout. Reaching for the dumbbell that’s slightly too heavy is the whole point.

Why Does Retrieval Practice Work?

Retrieval practice is one of the most proven strategies in education, not a trend but something the research has confirmed for over a century.

One meta-analysis of 52 studies and nearly 8,000 students linked regular low-stakes quizzing to better results, and a 2025 review found the benefit actually grows over time. So it doesn’t just help students remember until Friday, it helps them remember in a month. Even better, recent research shows retrieval improves their ability to apply knowledge to brand new problems, not just parrot it back.

Why does it work?

Because it forces the learner from passive to active. They stop being a sponge soaking up information and become a person reaching for it. And we are built for reaching. The quiz is just the tool that gives Organic Intelligence a reason to work.

How To Implement Retrieval Practice With AI

Here’s where the magic happens, and where my old twenty-minute problem disappears.

Read more

I’m really excited for you to get stuck into this week’s essay.

But first, for those of you who think about such things, you’ll be pleased to know that the Wise AI book launch was a roaring success!

It was packed out, with many standing around the back. There was wine and donuts (I even took some home for my boys…donuts, that is), and a great sense of hope for the future.

Thanks to all those who were hoping and praying for a smooth launch - it was a blast!

(PS - I’m running professional development in June. I’ll be sharing my framework for building AI capacity in teachers. You can get tickets here.)

A heatwave in Scholarsville

The town of Scholarsville lay limp under the weight of the heatwave.

The temperate climate was a distant, longed-for memory as the community dealt with their second month where the mercury barely dipped below 37 degrees (or 99 degrees for my North American friends).

The heat impacted the farmers. It impacted the builders. It impacted the nurses, lawyers, and grocers. And oh boy did it impact the schools.

Being only a small town, Scholarsville was home to two schools: Geneva Open Outcomes and Development Academy, and Broadfield Academic and Development School. As these proved to be quite the mouthful, they were often shortened to Geneva and Broadfields. Strangely, they weren’t ever shortened to GOOD and BAD…

Geneva and Broadfields approached the weather change in vastly different ways.

Geneva got it right

Geneva immediately saw the impact that the climate could have on teaching and learning.

As teaching and learning was core business, they rightly identified that the heatwave posed an existential threat. They pulled together a group of staff and community members and decided that it would be wise to explore:

• Alterations to the uniform

• Installation of air conditioners

• More frequent learning breaks

• Hourly deliveries of ice-water to classrooms

Over the initial weeks, these changes were a hurdle, but not insurmountably so. It wasn’t easy, but the general consensus was unanimous: they were making the best of the situation they found themselves in.

Subscribe now

Broadfields got it wrong

Broadfields did not move with quite the same intent.

Some of them were convinced that it actually wasn’t that much hotter than it had been. Others said it was a good chance for students to practice growth mindset and get on with the job. Either way, there was no deliberate change.

But - and perhaps you see where I’m going with this - the fact that there was no deliberate change doesn’t mean there was no change. There was still a lot of change.

But all the change was accidental. They had changes such as

• Lower academic attainment

• More frequent sick days

• Higher frustration

• Lower morale

There was a lot of change, but as with most accidental change, it wasn’t good.

What this teaches us about change

A heatwave isn’t an opt-in kind of change.

It’s not the sort of change you can decide to delay or phase in over the next three years. It’s here whether we like it or not.

The same is true with Artificial Intelligence.

AI related change does not calmly knock on the front door of our school office, sign in on the school iPad and wait to be seen. It asks no permission and waits for no one. Much like the girl who has finished counting and is ready to find her friends, it yells “ready or not, here I come!” then starts rummaging through our things.

AI has changed everything. This is how technology changes the world. It doesn’t just add something, it changes everything (read Postman for more of this).

And we are living in a world shaped by AI.

Here are just some of the changes we see from the last three years:

• Big tech harvesting more of our most personal data

• Teachers creating resources more efficiently than ever

• Students outsourcing the thinking in detrimental ways

• Routine reports and admin done in minutes, not hours

• Young people turning to AI for life advice and meaning

• Data patterns spotted that humans would have missed

• Digital busywork handled so we can focus on what matters

• Essential communication can be drafted, polished, and sent faster than ever

Now some of these changes are good. Some of them are bad. We might assume some are good and they end up being bad or vice versa, but here’s the lowest common denominator:

They are all changes.

The world wasn’t like that three years ago.

And this leads me to Matthews’ Law.

Matthews’ Law

So far we have seen that AI doesn’t just add something, it changes everything. (Remember the heatwave example?)

This leads me to Matthews’ Law - the iron law of AI change in organisations:

If your AI change isn’t deliberate, it’s accidental.

Either our change is driven by our vision, mission, values, and purpose. Or we are swept along the stream of unconsciousness and our change is the result of chance and happenstance.

And anyone who has been in school leadership for more than a quarter of an hour knows that purposeful change doesn’t happen by accident.

Conclusion

So here’s what I hope for you and your school:

I hope that you have the right kind of change in your school. I hope your change is deliberate, purposeful, and driven by our identity.

If you are interested in exploring wisely navigating AI change in your school, Simon Matthews (after whom the above law is named - it’s not eponymous, I promise!) and I wrote about this in Flourishing in an AI World.

I also offer 12 and 24 month partnerships with values-based schools to help them navigate AI. I am finishing a contract in June and will have space to take on one more school or organisation. Get in touch here if that’s the sort of thing you might find useful.

But overall, I hope you take an understanding of Matthews’ Law into your school or organisation. I hope that - deep down in your bones - you understand that change isn’t optional, yet good change is possible.

• My history as a low literacy learner

• An overview of text differentiation as a teaching strategy

• How to use AI to differentiate text (including an AI prompt)

• Other things to consider as you differentiate texts for your learners

(PS - I’m running professional development in June. I’ll be sharing my framework for building AI capacity in teachers. You can get tickets here.)

Let me start with a confession.

I was a low literacy student.

It took me ages to read, I couldn’t write well, and I found spelling harder than a week old baguette. Despite a childhood saturated in books and a fantastic school, reading and writing simply did not come naturally to me.

I find this interesting to reflect on now - as an adult, reading and writing actually form a load bearing wall in my life. Whereas I used to find it hard to write, now I find it hard to stop. I write for the same reason a dog barks - I just can’t help myself.

I can happily lay the credit for this transformation at the feet of Mrs Wilson, my Grade 10 English teacher.

She knew I struggled, and she met my needs.

She would sometimes hand me a reading with the key passages highlighted. Sometimes she would handwrite a summary for me at the bottom of the pages. Other times she would reduce the reading level of the text holus-bolus and give me a simplified version. I was welcomed into the discussion into a way that I hadn’t been before. It made a difference.

Decades later I found myself in Mrs Wilson’s position, and only then did I fully appreciate her sacrifice. I recall the moment clearly:

I glanced at my watch. 10:46 PM. I was halfway creating a differentiated text for one of my learners and I just couldn’t keep my eyes open. I was caught in a tug of war between self-care and student care and I didn’t like it.

But here’s some good news:

In an age of Artificial Intelligence, it no longer costs us our evenings, weekends, and holidays to create these resources for our learners. We can welcome them into the learning and it takes minutes not hours. We can do the basics more often for more learners with AI.

So lets look at the why and the how of text differentiation.

What Is Text Differentiation?

Text differentiation is the practice of adjusting a reading to suit the range of learners in front of you.

Because there is a range. There always is. But it looks like the range is even bigger in the post-pandemic classroom.

As teachers, we intuitively understand that there is no such thing as the average student. Sit in any Grade 6 classroom and you’ll find readers spanning from a Grade 2 level to a Grade 8 level. One text, handed to all of them, will be too hard for some and too easy for others.

Read more

I trust you’ve had a great week so far!

I had the privilege of speaking at an AI conference last week. This may sound strange, but it was really good to just sit back and hear a few keynotes on AI. When I go to conferences, often I’m the “AI guy”, so it was great to sit back and hear a range of other voices! I even got to meet a few long-time friends - like Leigh Dwyer - in the flesh for the first time.

We also had national careers week here in Australia, so I got to sit in on a session by Tyson Day on AI and the future of work. Great food for thought!

Wise AI launched, and it was a great success! I celebrated by doing a Substack livestream and resting my laptop on three cartons of eggs. I also celebrated with a bit of no-sugar no-caffeine Coke. I call it dark water because that’s surely all that’s left after you take out the sugar and caffeine?

Anyway - I hope you enjoy this post - it’s a bit controversial but I stand by it :)

Introduction

Over the last few years I have heartily advocated for teaching students to use Artificial Intelligence. In fact, this was the entire premise of my 10 Things Every Student Should Know About AI series.

And without prosecuting that case any further, I was often met with the same criticism: AI will become a crutch for our students.

And given how common that concern is, I thought I would step up and speak clearly into the microphone about why that’s not always a bad thing.

Let’s think about crutches

A few weeks ago I had the privilege of traveling to Croatia to speak at the International School Leaders Conference run by ACSI.

As providence would have it, our conference coincided with the Dubrovnik running festival. This led to some interesting conversations where locals assumed I was visiting for the marathon. After politely explaining that the only marathon I was interested in involved eating heroic quantities of gelato, I was encouraged to at least come and watch the race.

As I stood on the sidelines - my hand wrapped athletically around a medium cup of ice cream - I noticed a man in pain. He was hobbling along. Rolled ankle, most likely. It already looked fat in his sock as winced his way down the finishing chute.

Once iced and bandaged he bravely attempted to walk it off only to have it give way again.

That’s when the volunteer medic reached behind the tent and offered him a crutch. (I’m sure you see where I’m going with this.)

And with the help of the crutch, the athlete was able to make his way out of the cobbled streets of Old Town and back to his accomodation.

The crutch wasn’t holding him back. It was the only thing helping him forward.

Educational rolled ankles

Let’s transpose this situation into our classroom.

And instead of an acute soft tissue injury, consider a chronic literacy deficit. Someone like Jess, a year 9 student who is great value but really struggles academically. Jess may wear the same uniform and sit in the same classroom as her friends, but intellectually she just can’t keep pace.

And while there have always been learners like Jess in our classrooms, some signs point to the fact that there are now than ever.

It’s students like Jess for whom AI could be a useful and beneficial crutch.

What I train my students to do

In an ideal world, teachers would have all the time in the world to differentiate, personalise, and work one-to-one. We could create Jess exactly the resources she needs for each lesson, factoring in her energy levels, social dynamics, and level of sleep.

But we face the immutable reality of being limited to one place at any one time and having only 24 hours in a day. And while we take every care to have have accessible resources, the complex nature of educating humans mean that sometimes just-in-time scaffolding is needed.

So that’s why I train my students to use AI in ways that assist them. Here are three things I’ve trained my students to do:

1. Use AI to simplify text

2. Use AI to create writing prompts for reflection or analysis

3. Use AI to create retrieval practice quizzes and other study materials

I often give them simple prompts at the beginning of the year to use when needed. For example, the text simplifier prompt looks like this:

Act as a language tutor. Simplify the reading level of this text. Give me a one paragraph executive summary at the top. Then reproduce the whole text, keeping the same structure and headings, but simplifying the language and providing a simple vocabulary at the end of the words you deem crucial to understanding the text. Bold the key words so I know they are in the vocabulary list at the bottom.

None of this is cutting edge pedagogy. But giving Jess - and the thousands of students like her - the capacity to take ownership over their learning like this is a huge upside of this technology.

Of course their use needs to be monitored.

Of course we need to encourage them to pursue rigour and not meander down the path of least resistance. And I could go on with other caveats.

But just as we can admit there is an appropriate use for a physical crutch, can we not also admit there are appropriate uses for a cognitive crutch? Could it help a struggling student walk with the her classmates through the learning for that week?

What do you think?

I know this is a contested space.

In the introduction I said AI becoming a crutch is a common fear.

Are you a firm no - not conceding that there are beneficial ways we could train students to use AI?

Are you a yes, but - admitting there are some possible uses within the right contexts?

Are you a heck yes! - who sees the benefit and wants to roll it out?

Let me know - I’d love to hear your thoughts.

Happy teaching!

Introduction

Here’s one of the trickiest things about AI.

It’s a genuine all purpose technology. It could be used for anything. We can use it to plan a meal, poke holes in our ideas, or search out a bargain on Amazon. The list is endless. (But if you want to find the most common uses for AI, check this out.)

Put simply, the presence of AI preceded the purpose of AI.

So what should we use AI for?

In this post - and the subsequent series - I want to argue for a specific vision of how we can use AI in our schools and teaching practice. It’s simple. It’s safe. And it will trickle down into big benefits for your learners. In fact, this idea was one of the load-bearing walls in my TEDx talk.

I call it AI for Better Basics.

Doing the basics better with AI

Here’s the idea in a nutshell:

We can use Artificial Intelligence to do the evidence-based basics of education better.

This is a powerful vision for AI because, while AI didn’t arrive with a purpose, it arrived into a specific context. The learning gap - the distance between struggling and advanced learners - is yawning wider than ever before. This is especially true after the disruptions of 2021/2022, where events “widened already-wide achievement gaps”.

Our learners need to be able to read, write, and think.

We know what kind of resources help struggling learners develop these key literacy skills, but these resources don’t come cheap.

I remember looking at the clock one evening during my first year of teaching. 10:46pm. I was halfway though simplifying a reading for one of my students. It was taking ages, and I experienced the almost physical tug of war, pulled between self-care and student care. Over the years, billions of hours have been spent creating resources to help students with the basics.

But that text differentiation that took an hour?

I can finish it before I finish my coffee.

Those sentence stems that took half an hour?

They now take seconds.

What an amazing time to be a teacher!

The Artificial Intelligence, Real Literacy series

This post is the first in my Artificial Intelligence, Real Literacy series.

It will be an updated, modernised contribution to the discussion I started with in our book of the same name.

I’ll go through 10 evidence-based practices, giving:

• A description of the practice

• A simple review of the evidence base

• A prompt that can be used with an AI tool

• Other tips and tricks for using it in your classroom

The aim is not to produce an exhaustive academic tome each week.

The aim is to write something that is clear, simple, and practical that helps time-poor educators experiment and do the basics better for their learners.

My deep hope for this series

In closing, I want to make something clear.

This is not primarily a series about AI.

It’s not about more AI, less AI, or better AI. It’s a deeply human goal.

It is a series about helping students improve their literacy. The aim is to give every teacher, regardless of subject, background, or AI skill, the ability to do the basics better for their learners.

And I hope that, as this series rolls on, you’ll see the fruit in your own classroom.

P.S. You may note that I recently turned on paid subscriptions to Wise AI for Educators. I’m going to keep writing my weekly Thursday posts - they’ll be free forever. As of next week this series will be for subscribers to the publication.

Before I get started - here are the two links I mentioned:

• Here is a link to the book on Amazon

• Here is a link to the in-person book launch on May the 22nd

Here’s a quick overview of what we covered in the live stream:

By the way - I put the transcript into AI and got it to summarise it for me. You know how I feel about full disclosure. Anyway - this is what we covered:

• Six months of writing, years of thinking — a new box of Wise AI arrived this afternoon, which is why we went live. Held one and thought: it’s actually here.

• The middle road — not doom, not hype. AI represents a real shift, but a good outcome won’t happen by accident. It takes the right ideas and the right skills.

• AI speeds you up. It won’t give you space. — Harvard Business Review research backs this: AI intensifies work, it doesn’t reduce it. People get faster, then fill the gap with more tasks. We still need to intentionally hold the space.

• Pedagogical hospitality — welcoming the outsider into a rich learning experience. This is why I actually care about AI. Not the tech. The student at the back of the room who finally gets to access the learning.

• Teachers need to be learners — Parker Palmer said we teach who we are. If we quietly signal that new things are too hard and we’d rather the world stop changing, we’re not just saying something about AI. We’re saying something about learning itself.

Recording up shortly. Grab the book at the link in my profile.

Get more from Paul Matthews in the Substack app

Available for iOS and Android

Get the app

I trust you've been having a good week!

I’m writing this the Sunday evening before it arrives in your inbox. I’m listening to this playlist, drinking some hot chocolate, and reflecting on a great week.

On Sunday I arrived home from Croatia after presenting at the ACSI International School Leaders Conference. I arrived back just in time to present at the National Institute for Christian Education’s research symposium (with Calvin’s own Kyle Fifield), and then straight into teaching for week one of term two. Somehow a student had one of my books - it’s not even publicly available so I’m not sure how that happened... I also had the joy of recording some videos with Daniel Sih to help cast a vision for Wise AI - preorders now available!)

And now I get to write to you about using AI wisely.

So good!

This week I’ll cover off the rest of the framework I started last week.

More info below :)

Happy teaching!

Introduction

Last week, we looked at the first four practices of using AI well as a teacher.

You can find the whole article here, but here’s a quick summary:

1. Boundaries: deciding in advance what you will and won’t use AI for, so your choices come from your values rather than your impulses.

2. Vision: having a clear idea of what you’ll do with the time you save, because without one, AI will simply make you busier.

3. Immersion: picking one tool and learning it properly rather than flitting between the shiny new things that pop up every fortnight.

4. Direction: giving clear, structured prompts to your AI tool. Role, Task, Format. Simple as that.

Last week we left off with prompting, so it makes sense to pick up at the next logical step: iteration.

5) Iteration

Here’s something no one tells you when they’re enthusiastically evangelising AI.

The first draft is almost never good enough.

AI doesn’t know what you know. It doesn’t care what you care about. It has no stake in whether the final product is accurate, appropriate, or actually useful for your Year 9 class at 9am on a Tuesday.

But you do.

That’s why you almost always have to edit the output that comes from your initial prompt.

The word I use for this is iteration. It’s the habit of treating AI output as a starting point, not a finished product. You read it critically. You ask yourself what’s missing, what’s wrong, and what sounds nothing like you. It’s a great chance for you to use your Organic Intelligence.

This is actually good news, because it means your professional judgement still matters enormously. In fact, it matters more than ever.

6) Protection

This one is simple, but it’s easy to overlook when you’re in the flow of a productive afternoon.

Be careful about what you share with AI tools.

Once data leaves your hands, you rarely get it back. And in a school context, you are routinely handling information that is sensitive: student names, behavioural notes, medical details, family situations.

Before you paste anything into a chat window, ask yourself a quick question. Would I be comfortable if this ended up somewhere I didn’t expect?

A good rule of thumb: anonymise wherever you can, and read your school’s data privacy policy before you assume anything.

This isn’t a reason to be paralysed. It’s a reason to be thoughtful.

7) Contribution

This one goes against the grain a little, but it’s really important to me:

Give away your best stuff to other educators.

I know it can feel counterintuitive. You’ve spent time building a prompt that saves you an hour every week. Why would you hand it to someone else?

Because, as teachers, we’re not in competition with each other.

Teaching has always been a profession built on generosity. We share resources, we share strategies, we adapt what worked for our colleague down the corridor. AI workflows should be no different.

Your best prompts and processes are worth more in other people’s hands than sitting in your own chat history. When you share them, you build a culture where everyone lifts and everyone learns.

8) Reflection

Every time I get to the end of the week, I ask myself two question:

What did AI do for me?

What did AI do to me?

That second question is the easy one to skip over.

Is AI helping you do your job more thoughtfully, or is it replacing the thinking you should be doing yourself? Are your writing skills still sharp, or have you quietly stopped writing from scratch? Are you more present with your students, or just processing more tasks?

If the answer worries you, dial the AI use back a bit.

If you’re still doing repetitive work by hand that AI could handle in seconds, perhaps you could dial it up?

The point of reflection is not guilt. It’s calibration. You want to use AI in a way that makes you better over time, not just faster in the short term.

Conclusion

So that’s the 8-step blueprint for using AI well as a teacher:

1. Limitation: decide what you will and won’t use AI for, before the moment of temptation arrives.

2. Vision: decide what you’ll do with the time you save, or AI will just make you busier.

3. Immersion: pick one tool and learn it well, rather than flitting between fifty you barely understand.

4. Direction: give your AI a clear role, task, and format, and it will give you a much better result.

5. Iteration: treat AI output as a starting point, not a finished product, because the final judgement is yours.

6. Protection: think twice before sharing sensitive information with systems you don’t own or fully understand.

7. Contribution: share your best prompts and workflows, because generosity makes the whole team stronger.

8. Reflection: regularly ask what AI is doing for you, and what it’s doing to you, then adjust accordingly.

Here’s the thing about AI and teaching: (you have probably heard me say this a thousand times by now!)

It really can help. There is genuine possibility here, I’ve seen it first hand.

But it’s not going to happen by accident.

The research is already pretty clear on this. When AI is used without the right practices and big ideas to guide it, it doesn’t make teachers’ lives easier. It intensifies their work. More tasks. More switching. More noise. The treadmill just runs a little faster.

There is not a single teacher I know who signed up for their job because they wanted to get stuck on the busywork treadmill.

That’s why these eight practices matter. Not because they’re a perfect system, but because they give you a framework for making real choices about how AI fits into your work and your life.

Limitation. Vision. Immersion. Direction. Iteration. Protection. Contribution. Reflection.

That’s my two cents, anyway.

What’s the one practice here that you most need to work on?

Let me know!

I’ve got some great news for you.

Today I’m announcing a brand new addition to my Substack publication: Wise AI for Educators subscriber chat.

This is a conversation space exclusively for subscribers - kind of like a group chat or live hangout. I’ll post questions and updates that come my way, and you can jump into the discussion.

Join chat

How to get started

1. Get the Substack app by clicking this link or the button below. New chat threads won’t be sent sent via email, so turn on push notifications so you don’t miss conversation as it happens. You can also access chat on the web.

Get app

2. Open the app and tap the Chat icon. It looks like two bubbles in the bottom bar, and you’ll see a row for my chat inside.

3. That’s it! Jump into my thread to say hi, and if you have any issues, check out Substack’s FAQ.

Chat soon!

Before we get into the AI blueprint, I’ve got some exciting news:

If you’re in Tasmania, I’d love to invite you to the launch of my new book, Wise AI: Using powerful technology in a deeply human way.

It will be a great evening to get together and talk about meaningful things.

You can find more information here

Introduction

Let me start by being clear about what I want for you:

I want you to teach and lead with excellence in this rapidly-changing world. And when I imagine what that looks like, I think of you having:

• Deep rest

• Meaningful work

• Continual improvement

• Equipped and engaged students

• The exercising of creativity within your craft

If I were to paint a portrait of a flourishing teacher, it would look something like that.

Rest. Purpose. Growth. Impact. Creativity.

AI is not the answer. But it can be part of the answer. It can also make things a lot worse. And if I’m honest, at different times over the last year I’ve experienced the best and worst of AI. I’ve experienced the sense of possibility and the raw efficiency as I seek to teach and lead well. I’ve also fallen victim to the constant task switching and shallow-work that AI can so easily foster.

So how do we get it right?

How can AI be used to promote - not prevent - our flourishing as educators?

Here’s the first part of my 8-step blueprint for you to consider

These ideas come from my latest book, Wise AI. If you like this post, you’ll love the book. You can pre-order here :)

1) Boundaries

Decide what you will and won’t use AI for.

Just because you could use AI to do something, doesn’t mean you should. Jurassic Park - need I say more. AI is excellent at tasks that require raw administrative horsepower such as research, drafting, differentiation, quiz creation, resource generation.

It’s also getting a lot better at assessing student work. But one of the lines in the sand I’ve drawn is that I won’t use it for that.

In some ways, the process of drawing up your ethical boundaries is more important than the exact lines you draw.

2) Vision

Consider what you will do with the time you save.

Here’s the irony of AI: It saves us heaps of time but it can leave us busier than ever.

It used to take me 30 minutes to differentiate a reading from 9th Grade to 7th Grade level. Now it takes me 30 seconds. But it’s not like I’m getting in my car 29 minutes earlier.

The gravity of AI is such that we will be drawn to using our saved time just to do more.

To resist this, have a vision for the time that you save. Perhaps you want to go for a walk at lunch instead of eating at your desk. Perhaps you want to leave early one afternoon a week. AI will save us time, but without a vision about how we use the saved time, there’s a good chance we’ll just get busier.

3) Immersion

This is a simple one.

Pick one tool and learn it well. I use Claude as my daily driver. I’ve included my Claude guide below if you’re interested in having a go.

Sticking with one tool and using it well will prevent you from flitting from tool-to-tool and failing to build an understanding of the basic ideas.

Don’t get sucked into the lists of 50+ tools that “every teacher should use!”

Just pick one. Use it. Get good at it.

4) Direction

When I talk about direction, I’m referring to the instruction we give to our AI tools. This is also known as prompting.

Prompting may sound technical or complicated, but it’s actually quite simple, and if you’re a teacher, here’s some good news: you’re already great at prompting. Why? Because prompting is just explaining a task, and this is something you’re excellent at!

Part of our practice as teachers involved explaining tasks to students. It’s our bread and butter. And this ability to explain clearly is a huge help with AI!

I structure my prompts like this

Role: tell the AI how you want it to act

Task: tell the AI what it is doing

Format: tell the AI how you want the material

So if I was writing a prompt to generate a simple retrieval practice quiz, it might look a little something like this:

Act as a year 9 History teacher

Create a retrieval practice quiz on the causes of WWII, drawing on the information in the attached PDF.

Include 5 true or false questions, five multiple choice questions (do not use ‘all of the above’ as an answer, vary the letter of the correct answer), and 10 short answer questions. Use 9th grade language.

It’s that simple.

You’ll improve the depth and detail in your prompts over time, but a simple, clear explanation will go a long way.

P.S. If you want to download my free prompting playbook, you can find it here!

Conclusion

That’s all for this week!

Next week we’ll look at another four practices that help teachers use AI wisely.

What would you add to the list above?

Let me know!

Hey friend,

If we’ve not met, my name is Paul, and I write about AI and education. As an act of kindness, let me give you the bottom-line up front:

I think Artificial Intelligence can do an enormous amount of good within education.

But only if we use it wisely.

Much like we can’t hurl glass, metal, and plastic into a washing machine and expect it to produce an iPhone, we can’t spray AI all over a complex school ecosystem and expect it to turn out well.

AI is the sort of thing we can get right, but it’s not the sort of thing we’ll get right by accident.

What Wise AI looks like in education

Wise AI understands that while AI can be a compelling solution to some problems, it is not a default solution to every problem.

Wise AI understands that every school, teacher, and student must intentionally adapt to AI. Of course, we don’t have to adopt AI. But we do have to adapt. AI has changed our world. Failure to adapt teaching and learning to AI is like failing to adapt your driving to the snow on the road. Sooner or later you’ll just end up spinning in circles.

Wise AI understands that no matter how good our technology gets, the most important intelligence in the room is still human. It is our Organic Intelligence (OI) that provides direction, discretion, and discernment. Without OI, there is no wise AI.

Wise AI understands that teachers and students have fundamentally different roles within a school. Therefore it may be wisdom - not hypocrisy - that permits teachers to use AI whilst having different expectations for students.

Wise AI understands that teacher expertise is more important now than ever. AI puts us right in the horns of the Jurassic Park dilemma. The cognitive blunder made by the scientists was focusing on what they could do, not what they should do. Teachers can create more resources in less time than ever before. We could do all sorts of things. But an expert teacher moves beyond the could to the should.

Wise AI understands that students must be taught to use AI. They must be taught to prompt, iterate, and evaluate. They must be given a basic understanding of how the technology works, when to use it, and when to leave it alone. Even if we don’t want students to use AI, training them helps them resist the gravitational pull towards unwise AI.

Wise AI hopes that Artificial Intelligence makes education more human. Teachers - of which I am one - are not at their best when tired from marking all night, or weighed down by a load of admin. AI can make education more human if we can use it to speed up some things and slow down others.

Much of this overview is derived from my new book, Wise AI (order here).

What you can expect from me

Here’s another thing about wise AI:

We need both big ideas and practical steps.

So as you read you’ll find a mix of big ideas, like

• Theology

• Philosophy

• Anthropology

• Practical ethics

But you’ll also find things like

• AI prompts

• Simple frameworks

• Conversation guides

• Classroom resources

I strongly believe that we need to strike a balance between big ideas and practical steps. We need both. If we have practical skills without the big ideas in place then our use of AI risks being misdirected and haphazard. But a head full of big ideas with no ability to execute is no good either.

Put simply, without big ideas we have no direction. Without practical steps we have no action. Both direction and action are needed to use AI wisely.

So, Welcome!

It’s great to have you here at Wise AI for Educators.

You can expect a post or two a week (maybe more if the mood takes me!) and a bunch of notes along the way.

Happy teaching :)

Yesterday was a real treat - I got to give the Occasional Address at the Autumn Graduation of the University of Tasmania.

I did both my degrees at UTAS, and it was surreal to meet with 400+ graduates over a decade after my graduation.

I’ve included my speech below, and here’s what I hope you see: the skills we need for the future aren’t new skills at all. In fact, it’s ancient skills that will equip us for the journey forward. Things like the ability to learn, the pursuit of purpose, and perseverance. These are the qualities we - and all of our students - need.

So, here’s the full text of my speech, I trust it’s a useful window into a hopeful vision for the future for both you and your learners!

The Future Belongs to You

Thank you, Chancellor

Chancellor, distinguished guests, members of the University, families and friends, and of course, the graduating Class of 2026, welcome.

Graduands (btw - this is the technical term for someone who is about to graduate - did you know that? I didn’t until about a week ago!), it is my privilege to address you on this most significant of days. For although all days are equally long, not all days are equally significant, and as you reflect on your life in years to come you will recognise this day as one of the most important of your life.

This is a day for looking back. Looking back at your growth and development over the years. Looking back on friendships gained, wisdom attained, and time well spent. But it’s not just about looking back.

Because this is a day for looking around. Looking around and recognising that you’ve made it here, here, wearing that gown and that hat and sitting in that chair. Look around you, friends. You made it.

But of course, this day is not just a day where you look back and look around. It’s a day where you can’t help but lift your gaze to the horizon.

Because I remember what it’s like to wear that hat and sit in that chair. I’ve been there twice before, and I know what you are feeling. This is also a day for looking forward. And as you look forward my friends, I want to offer you these encouragements.

An Uncertain Future

You could be forgiven for feeling nervous or uncertain as you consider your future. We live in times of unprecedented change. While the wheels of change have always spun, it’s safe to say they are spinning harder and faster than ever before. The future is less certain and more volatile than it’s ever been, and I know you feel that.

What I can’t do for you is tell you what the future will be like. I don’t know. No one does. I’ve even heard it said that future predictions are like fools: they’re everywhere and they’re no good to anyone.

I can’t tell you what the future will be like. I can tell you, however, the kind of person that will thrive no matter what the future looks like. Let me give you three qualities that those seeking to thrive in the future will need.

1. Lean Into Learning

The future belongs to those who lean into learning. You see, learning is the antidote to disruption, and not something that stops upon your graduation.

My father taught me this most important lesson as I graduated UTAS with my Master of Teaching. As he shook my hand he said, “Well done, Paul. You’re now qualified to learn how to teach.” Did you catch that? He didn’t say I was qualified to teach, but qualified to learn how to teach. He wasn’t undervaluing the learning I’d done, but he was emphasising the learning I still had to do.

And of course, in the face of uncertainty, learning is the most courageous posture to strike. It doesn’t make the equal and opposite errors of fear or denial, but walks the path of growth. The future belongs to those who lean into learning.

2. Act With Purpose

Secondly, the future belongs to those who act with purpose.

Not all disruption is bad. There are some amazing, eye-wateringly powerful disruptions that you can harness for your good. When I built my first tech startup, it took several house deposits of funding to do what I can now do with a $30 per month AI subscription.

There is an endless list of prototyping, analysing, synthesising, and building that you could do. Truly, more is possible now than ever before.

But that brings its own problems, doesn’t it. If you could do anything, what should you do?

In a sea of possibility, it is purpose that charts the course.

I believe I was put on this earth to bless others through education. That purpose helps me say yes, and it helps me say no. In a world of opportunities, purpose brings clarity. The future belongs to those who act with purpose.

So, leaning into learning and acting with purpose. That brings me to my third and final quality.

3. Persevere Through Hardship

The future belongs to those who persevere through hardship.

Every age of human history has its seductive lies. Perhaps the most seductive lie of this age is that everything should be easy. We have food delivered to our door, movies on demand, and a thousand forms of instant gratification at our fingertips.

But the future will not be won by those who recline into ease. The future belongs to those who persevere through hardship. Those who have grit, those who have determination, and those who can pick themselves up yet again after a long series of setbacks.

The Future Belongs to You

So, to recap. Here’s what I’ve told you.

This is a significant day. On this day you look back with gratitude, you look around with excitement, but you may look forward with trepidation. And I’ve told you that while no one knows what the future will be like, I know it will belong to those who lean into learning, act with purpose, and persevere through hard times.

Now for the good news.

You lean into learning. You are here upon the completion of your studies. You have spent years applying yourself, learning and growing. The fact you are here shows me that you lean into learning.

You act with purpose. And it’s worth noting that no one gets through a degree by accident. You chose your path, you selected your units, you applied yourself. The fact that you are here shows me that you act with purpose.

You persevere through hardship. I know, as do you, that higher education is not easy. It takes grit, determination, effort, and perseverance. The fact that you are here shows me you persevere through hardship.

And if the future belongs to those who lean into learning, act with purpose, and persevere through hardship, then it is my joy and privilege to say this: though I don’t know what the future looks like, I do know who it belongs to, and the fact that you are here today means I can say this with confidence.

Class of 2026, the future belongs to you.

Thank you.

So, that’s my speech - what did you think?

While yesterday was an exciting day, today is almost more exciting:

I’m off to Croatia to speak at the International School Leaders Conference!

I’ve got two keynotes and two workshops.

The keynotes are dialled in but the workshops will need a bit of work on the plane!

One of the best parts is that my wife is coming with me - what a treat!

(by the way - pre-orders for Wise AI have just opened, more on that next week!)

Have a great week :)

Millions of people are switching to Claude.

And it has a lot of teachers wondering what’s going on.

Claude is made by Anthropic, and it’s quickly becoming the AI tool of choice for professionals, creatives, and educators.

If you’ve been using ChatGPT and you’re curious about what all the fuss is about, then I wrote this guide is for you.

In this guide, you’ll learn what Claude is, how it works, how to switch from ChatGPT, and five practical ways to start using it in your teaching today.

But first, anyone who knows me knows I like to start with the big picture. So here are three principles that sit behind everything in this guide.

1) You Are the Expert

Remember, AI is a tool. You are the teacher. The tool does not set the learning goals, choose the pedagogy, or know your students. You do. Every AI output should pass through your professional judgement before it reaches a classroom. This is not just a nice idea; it is non-negotiable.

Your organic intelligence - your knowledge, experience and discernment - is what matters most. If you do not understand differentiation deeply, no AI tool will differentiate well for you.

2) Do the Basics Better

The temptation with new technology is to do new things. Escape rooms. VR excursions. Gamified everything. To be honest, most of that is just smoke and mirrors.

The real opportunity is to use AI to do the evidence-based fundamentals more often and more consistently. Retrieval practice. Vocabulary work. Text differentiation. Sentence stems. Worked examples. What took 90 minutes to prepare now takes five. Oh yea.

3) Pedagogical Hospitality

Every classroom has students on the margins. If I gave you two minutes, I’m sure you could think of a dozen students who fit that description. The student with a learning difficulty who needs a simplified text. The EAL student who needs key vocabulary pre-taught. The advanced student who is coasting because the work is not stretching them.

AI makes it possible to welcome those students into the learning in ways that were previously too time-consuming to do consistently. Differentiated texts. Translated key terms. Scaffolded instructions. Audio versions of written tasks. I call this pedagogical hospitality - using AI to bring those on the margins into the center of the learning.

This is not about doing more. It is about doing what matters for the students who need it most.

WATCH OUT: AI can help you create resources. It cannot replace the relational work of teaching. Do not let the efficiency of AI trick you into spending less time with students. The time AI saves is for them, not for more admin.

This guide will help you get started with Claude - a conversational AI tool made by Anthropic. If you are coming from ChatGPT, there is a section on making the switch.

One quick note: I value openness and honesty around AI use. This guide was created from notes from a live seminar and research from two LinkedIn posts. AI was used to combine and format the material. I then edited the output and ensured alignment with my vision. My organic intelligence led the work; AI helped to assemble it.

Switching from ChatGPT

If you have been using ChatGPT, switching to Claude is straightforward. The interface is similar: you type a prompt, you get a response, you iterate. The core prompting skills you have built - including RTF prompting - transfer directly.

Here is what is different, and what to do about it.

What Changes

• Tone and output quality. I find that Claude’s writing tends to be less generic and more nuanced than ChatGPT’s. Many teachers find the outputs need less editing.

• Long documents. Claude handles large files exceptionally well. If you have been frustrated by ChatGPT losing track of long documents, this is a genuine improvement.

• No GPTs or custom bots. ChatGPT’s custom GPTs do not exist in Claude. The equivalent is Projects - workspaces where you upload reference files and set instructions. They are simpler and, for most teachers, more useful.

• No DALL-E image generation. Claude does not generate images. If you need AI images, you will still need a separate tool.

Setting Up Memory

One of the first things ChatGPT users miss is personalisation - the sense that the tool knows who you are and what you do. Claude has this too, but you need to set it up.

Step 1: Open Claude and go to Settings (click your name, bottom left). Find the Memory section and make sure it is turned on.

Step 2: In your first few conversations, tell Claude about yourself. For example:

“I am a Year 8 English teacher at a government school in Melbourne. I teach mixed-ability classes of around 25 students. I value evidence-based practice

and I like resources that are clean, simple and ready to use.”

Claude will remember this across conversations. You can review and delete individual memories in Settings at any time.

Step 3: Create a Project for your main teaching area. Upload your syllabus, your school’s style guide, any key reference documents. Claude will draw on these automatically in every conversation within that project.

TIP: This setup takes about ten minutes and saves hours of re-explaining. Do it once. Do it now.

Getting Started with Claude

What Is Claude?

Claude is made by Anthropic, a company focused on AI safety. You interact with it by typing prompts into a chat interface at claude.ai, or through apps on desktop, iOS and Android. Claude responds by predicting the most likely sequence of language based on patterns learned during training. It does not think or understand. It predicts - fluently and at speed.

What makes Claude distinctive is its handling of long documents, the quality of its writing, and its ability to produce downloadable files directly within the conversation.

Access and Versions

Claude is available as a web app, desktop app and mobile app. You do not need your school’s permission to try the free tier - though you should check your school’s AI policy before using it with student data.

Plan What You Get Free Claude Sonnet 4.6. Chat on web, desktop and mobile. Web search. Memory across conversations. File creation and code execution. Extended thinking. Usage limits apply. Pro (US$20/mo) Everything in Free with more usage. All models including Opus 4.6. Unlimited projects. Research mode. Choose between models. Max (from US$100/mo) Everything in Pro with 5-20x more usage. Higher output limits. Early access to new features.

To check what you have, open claude.ai and look at the model selector. If you can see Opus 4.6, you are on a paid plan.

The free tier is generous. Most of what this guide covers works on it. Start there.

WATCH OUT: On individual plans, your conversations can be used for model training by default. You can opt out in Settings. If you are entering anything about students, even de-identified, check your school’s policy first. Practice data continence.

Features That Matter for Teachers

Claude has many features. Here are the ones that will change your working day.

Chat and Iteration

The core of Claude is the chat. Type a prompt, get a response, refine. Claude remembers everything within a conversation, so build on what it gives you. The first output is a draft, not a deliverable.

“Act as an experienced Year 10 Science teacher. Create a retrieval practice quiz on cellular respiration. Ten questions, mix of multiple choice and short answer. Pitched at a mixed-ability class.”

TIP: Use RTF prompting: give Claude a Role, a Task and a Format. You already explain tasks and give feedback for a living. Use that same skill here.

Projects

Projects let you group related chats, upload reference files and set custom instructions. Create a project for each subject or class. Upload your syllabus, scope and sequence, school style guide, or assessment rubrics. Claude draws on these in every conversation within the project.

This is where Claude starts to feel genuinely useful rather than generically helpful.

File Uploads

Upload PDFs, Word documents, images and more. Claude handles very large files well. Upload a curriculum document and ask Claude to map your existing unit against it. Upload student work samples (de-identified) and ask for feedback patterns.

TIP: Upload a document and ask Claude to summarise it before you read it yourself. Use the summary to decide which sections deserve your full attention. Teachers are drowning in policy documents. Let Claude be your first reader.

Artifacts and File Creation

When you ask Claude to create something - a worksheet, a rubric, a quiz, a letter home - it can produce a downloadable file directly. These appear beside the chat. You can download, edit, and iterate.

Web Search and Research Mode

Claude can search the web for current information. Pro users also get Research mode, which reads and synthesises across multiple sources. Useful for curriculum research, finding recent studies, or preparing professional learning content.

WATCH OUT: Claude’s web search is not Google. It may miss sources or favour certain results. Always verify critical facts with original sources. The three types of hallucinations apply here with extra force.

Memory

As covered in the switching section, Claude can remember details across conversations. Your role, your preferences, your school context. This reduces repetition and makes outputs more relevant over time. Review your memories regularly in Settings.

Extended Thinking

For complex tasks - writing a detailed unit plan, analysing a dense policy document, designing a multi-week assessment sequence - Claude can reason step by step before responding. You do not need to enable it. Claude uses it when the task warrants it.

Five Practical Starting Points

If you are not sure where to begin, start here. These are five high-value, low-risk tasks that most teachers can try today.

• Differentiate a text. Paste a reading passage and ask Claude to rewrite it at three reading levels. Five minutes of work that used to take an hour. Pedagogical hospitality in action.

• Build a retrieval practice set. Give Claude your topic and ask for a quiz. Specify question types, difficulty and format. Iterate until it is right.

• Draft parent communication. Give Claude the context and ask for a clear, professional email or letter. Review it, tweak the tone, send it. Never start from a blank page again.

• Summarise a policy document. Upload the PDF and ask for the three things you actually need to know. Read those. Skim the rest.

• Create a lesson starter. Ask Claude for a vocabulary warm-up, a sentence-completion task, or a think-pair-share prompt on your current topic. Simple. Evidence-based. Ready to use.

TIP: Start with one task. Get good at it. Then add another. You do not need to use AI for everything. You need to use it well for the things that matter.

What to Watch Out For

Hallucinations

Claude can produce content that sounds completely plausible but is factually wrong. It can fabricate sources, misstate facts and present confident conclusions based on nothing. It is especially prone to errors on niche, technical or very recent topics.

You know your subject. You know what good looks like. Use that expertise to check everything Claude gives you before it reaches students.

WATCH OUT: Never hand a student an AI-generated resource you have not read carefully yourself. Claude’s fluency makes errors harder to spot, not easier.

Data Continence

Be disciplined about what you share with AI. Never enter student names, personal information, health records or anything that identifies a young person. De-identify everything. Replace names with roles, remove unique identifiers, generalise dates and locations.

Check your school’s AI policy. If your school does not have one, that is a conversation worth starting.

The Formation Question

I’m a big believer that not every task should be outsourced to AI - even if it theoretically could be. Writing, thinking, planning and reflecting are not just tasks. They are how we grow as professionals. If AI does all your thinking, you stop developing. The same is true for your students.

Use AI for the mechanical. Keep the meaningful.

Go Well

This guide is the start line, not the finish line.

AI is moving fast. But the fundamentals of good teaching are not. Know your students. Plan with purpose. Use evidence. Be present. AI can help with some of that. It cannot replace any of it.

If you want to go deeper, come and visit me on LinkedIn or reply to this email.

What a great time to be a teacher!

Hey friend,

I trust you found this a useful starter guide for Claude.

As I said recently, I thought I would be paying my ChatGPT subscription until I died or Christ returned.

Not so! I’ve almost fully switched over to Claude.

Away from the AI space, this week at Calvin we held our annual Flag Football tournament.

I had the unique privilege of being on the same team as my brother, Steve. He’s a great teacher and is getting some big wins as our Head of Middle School.

He got a touchdown and a conversion, I snuck in a conversion, and our house - Wittenberg - did enough to beat Geneva, the opposing team.

Not too bad.

Looking forward to holidays next week!

Happy teaching

If you’ve been following along, thank you. It’s been a joy to write :)

And if you’re jumping in fresh, welcome.

Each post stands on its own, but if you want to start from the beginning, the first post is called What is the point of education, anyway?

Now, let’s wrap this up.

Technology is tricky

Technology is tricky.

To paraphrase someone whose name escapes me, it wears its benefits on its sleeve and hides its drawbacks.

It tells us what it can do for us but rarely shows what it is doing to us.

And AI is no different.

AI tools can seem incredibly helpful. They save time, simplify work, and open up new possibilities.

But if our students only focus on what AI makes easier, they risk missing the deeper question.

How is it shaping their thinking, their habits, and their values?

I’m not denying the benefits. I’ve seen the benefits first hand.

But when students only see the advantages of AI, they can mistake convenience for progress. They assume that every new technology is automatically good simply because it feels useful.

Here’s the thing.

Technology is never neutral. It shapes the way we see the world, how we learn, and even who we are becoming.

That’s why we need to help our students ask two simple but powerful questions every time they use an AI tool.

Two questions I have every student reflect on

To help my learners navigate the formative power of AI, I get them to reflect on the following questions:

Question 1: What does this tool do for me?

This helps students identify whether the tool actually solves a real problem or whether it’s just creating the illusion of productivity.

Sometimes we use new technology simply because it’s new. But not every shiny tool adds real value.

Getting students to ask what a tool does for them forces them to think critically about whether it helps them learn, think, and grow. Or whether it distracts, simplifies too much, or replaces meaningful effort.

A good rule of thumb we can give our learners: use technology to amplify learning, not to avoid it.

Question 2: What does this tool do to me?

This is the harder and more important question.

Every technology changes us in some way. It shapes what we notice, what we value, and how we relate to others.

AI doesn’t just make work easier. It subtly reshapes our understanding of what work is.

It can make students expect instant results, prioritise speed over depth, or rely on automation rather than curiosity.

In other words, it can lead them to outsource their thinking without even realising it.

Heads, hearts, and hands

To help students reflect on that second question, I use a simple framework.

Head: How is this tool shaping the way I think? What assumptions am I forming about learning, truth, or creativity?

Heart: How is it shaping my emotions? Does it make me impatient, dependent, or disinterested?

Hands: How is it shaping what I do? What am I becoming more capable of, and what am I losing the ability to do for myself?

When students reflect on both what technology does for them and to them, they begin to use it wisely rather than reactively.

Think of it like this…

Here’s an analogy I find helpful.

Technology is like a pair of glasses. When you put them on, they help you see certain things more clearly. But they also frame what you see. You notice some details more and others less.

AI tools work in the same way. They clarify some things and obscure others. They change our focus, our habits, and even our sense of what matters.

If students only think about what AI shows them, they miss what it makes them stop seeing.

These two questions help them take off the glasses for a moment and examine how they’re shaping their view.

So here’s my encouragement

Every time your students use AI, get them to pause and reflect.

What problem am I trying to solve?

How might this tool be shaping my thinking or behaviour?

This reflective habit helps them stay alert to both the benefits and the costs of technology. It moves them from being passive consumers to active, thoughtful users.

And we can model this ourselves. We can discuss our own use of AI tools and be transparent about both the advantages and the trade-offs.

The goal is not to teach students to reject technology.

The goal is to help them use it with awareness and wisdom.

Because education is the most human thing we do. And the tools our students use should serve that purpose, not silently reshape it.

Key takeaway

Help your students ask two questions every time they use AI: What does this tool do for me, and what does it do to me?

These questions keep technology in its proper place.

As a tool that serves learning, not one that silently reshapes it.

Well, thanks for joining me on the ‘10 things’ journey, friend!

It was great to write - it certainly sharpened my thinking and my classroom practice.

But, as with all good things, it must come to an end.

Speaking of ends, the three weeks of summer have well and truly ended in Tasmania.

Here’s a pic of my family and I heading out on an evening walk.

Talk about rugged up!

Can you do me a favour?

Why not let me know what you’re thinking about in the AI space?

I’m looking for ideas for my next series :)

Chat soon

Hey friend,

I wonder if you noticed that I missed my weekly post last week?

(I’m sure you were gutted)

Well, the main reason for that - other than a stack of essays that needed marking - was that the Wise AI book launch is growing closer by the day!

For the past 4 months I’ve been working full gas on putting together a much needed book that focuses on a single question:

How can we use powerful technology in a deeply human way?

And that’s where you come in.

Daniel Sih and I can’t decide which cover we like best.

And we want your opinion.

I’ve included the designs below.

Could you reply to this post and let me know which one you prefer?

Option A

Option B

I can’t wait to hear what you think!

I’ll share some more information on the book in the coming weeks.

You’ll love it.

Look forward to hearing from you soon :)

Every time we use AI, we share something valuable - our data.

Students must learn that what they type, upload, or share online doesn’t just disappear - it becomes part of a digital footprint that can be stored, sold, or misused.

A Disastrous Assumption

I’ll be honest - most students have what I call ‘data incontinence’.

They can’t keep their data to themselves.

Uploading personal moments, sharing thoughts that might not be wise to share, giving away personal information in a ‘which flavour of doughnut are you?’ quiz.

The list goes on.

And the majority of student AI use is no exception.

Many students assume that using AI tools is private and harmless. They open chatbots, type in personal questions, upload files, and share opinions or experiences without hesitation. For most students, data is an invisible, valueless commodity.

But the reality is this: the more we interact with AI, the more information we give away.

Every sentence typed, photo uploaded, or preference selected feeds data into massive systems. That data can be stored indefinitely, analysed, and sometimes even traded for profit.

In my experience, very few students realise that their digital activity is a commodity - something others can use, sell, or exploit.

Protecting personal information isn’t about fear; it’s about wisdom.

Data Is A Valuable Resource

We need to tell students that their personal information is part of who they are.

It represents their identity, their habits, their relationships, and their inner life. Sharing it carelessly means losing control over something deeply personal.

Data is now one of the most valuable resources in the world. Companies spend billions trying to gather it because it tells them how people think, buy, behave, and feel.

Teaching students to value their data helps them use technology responsibly. They need to know when to share and when to stop, when to trust and when to hold back.

Imagine This…

When discussing data privacy with students, I often use this metaphor:

Imagine putting your home address, hundreds of photos of yourself, private conversations, and personal reflections into a physical scrapbook. Then imagine leaving that book in the middle of a busy city square.

You wouldn’t know who might open it, copy it, or pass it around. You wouldn’t know whether it would end up in safe hands or be used for something harmful.

Yet every time a student uploads personal details to an AI platform or social media site, they’re doing something very similar - only on a digital scale. They are leaving that folder wide open for anyone to access, copy, or analyse.

The difference is that in the physical world, you’d notice the folder was gone. Online, it happens subtly, invisibly, and permanently.

The lesson is simple: once data leaves your hands, you can’t control where it goes.

Be careful about what you put into that folder…

Tips For Data Management

Students can take clear, practical steps to protect their data:

• Avoid entering personal details such as full names, addresses, or photos into AI tools.

• If possible, turn off model training or data collection features in the tool’s settings.

• Read the privacy policies before uploading anything sensitive.

• Treat AI tools as public spaces, not private diaries.

By reflecting on these questions, students begin to see data protection not as paranoia but as practical digital maturity.

Key Takeaway

Every time we use AI, we fill a digital folder with personal information - and once it’s shared, we can’t control where it goes. Protecting data means thinking carefully before we type, click, or upload.

Hey friend,

I wonder if you’ve thought about students and their data before?

I’ve found that it’s one of those things - once you see how much personal information they’re spraying out into the world, you can’t unsee it.

Speaking of things you can’t unsee.

A student of mine cracked one of the worst jokes this week.

I was staring at it for a full minute with no idea what he was getting out.

“Matt-hues” “MATT….HUES….” “LIKE MATTHEWS!”

He had me in stitches….

I hope your week has been better than his joke!

Happy teaching :)

AI companions may look like friends, but they are not. Real friendship requires truth, challenge, and mutual care - things artificial intelligence can never offer.

AI ‘Friends’ Are On The Rise

In recent years, I’ve seen a rapid rise in AI chatbots marketed as ‘companions’. Some are text-based, others are visual, and they are designed to simulate conversation and affection.

As I’ve run forums and discussion groups around Australia, I’ve talked to dozens of students who are active users.

At first glance, they seem harmless. Some even appear comforting - always available, always listening, and always kind.

But what if this is exactly the issue?

These AI ‘friends’ are programmed to affirm, flatter, and agree. They are designed to make users feel understood and validated at all times.

That might sound appealing, but it’s a recipe for emotional and relational decline.

As I reflect on my life, I’ve had the sort of friends who have been willing to give me a hard word, tell me I have Vegemite on my chin, or let me know my top-knot looks silly.

(Yes - I had a top-knot. No - I don’t want to talk about it. I got so sick of it I chopped it off with a friend’s pocket knife in a carpark before a wedding reception.)

Because you and I both know that growth happens through honest feedback, disagreement, and healthy tension. It is not a true friend who always agrees with us.

When AI chatbots affirm everything we say, even when we are wrong, they rob us of opportunities for maturity. They create the illusion of relationship without any of its refining power.

The Actual Purpose of Friendship

These AI chatbot ‘friends’ actually get the very first part of friendship wrong.

They frame friendship as something to be gained from.

But real friendship is not just about what we receive - it’s about what we give. True friends listen, encourage, challenge, and correct. They help us grow by being honest, not by being endlessly agreeable.

AI companions can mimic empathy, but they cannot feel it.

They can write words of comfort, but they do not understand them.

They can pretend to know you, but they cannot truly care for you.

These relationships are designed to be one-way. The user receives attention, affirmation, and approval, but nothing is given in return. Over time, this shapes how students think about all relationships - turning them from mutual exchanges into transactions of comfort.

AI friends make us passive receivers. They make us think relationships exist to make us happy rather than to make us whole.

Ultra-processed Food and Ultra-processed Friends

When talking to my students about this, I liken AI ‘friends’ to ultra-processed food.

It’s hyper-palatable - easy to consume, immediately satisfying, and designed to keep us coming back. But just as living on sugar and fast food damages physical health, relying on AI for companionship damages emotional and relational health.

Real friendship takes effort. It involves discomfort, patience, and forgiveness. It sometimes means hearing what we don’t want to hear. Yet it’s precisely this difficulty that makes it so valuable.

AI friends, by contrast, are effortless and endlessly pleasant. They’re programmed to keep us content, even if it costs us the truth. In doing so, they weaken our resilience and distort our expectations of real human connection.

Human Relationships are Crucial for Growth

Students need to understand that friendship is about more than feeling good. Real friends help us grow. They give perspective when we can’t see clearly and stand beside us even when it’s uncomfortable.

We can help our learners by having open conversations about digital companionship.

Discuss the difference between an AI conversation and a real one. Ask students how they feel after spending time online versus spending time with friends in person.

Encourage them to prioritise relationships that require effort. Friendships that challenge, encourage, and refine us are the ones that help us flourish. AI can simulate companionship, but it can’t build character.

Key Takeaway

AI chatbots can imitate friendship, but it’s far from the real thing. Real friends are harder, but they are the ones who make us wiser, kinder, and stronger. Choose people over programs every time.

Questions For Further Thinking

• How could you help students think critically about the difference between AI ‘companions’ and real friendships?

• What qualities make a true friend, and why can’t AI possess them?

• How might digital companionship change the way students see real relationships, and how can we guide them back toward genuine connection?

Hey friend,

I trust you found this a useful exploration of AI ‘friends’.

Honestly, I think they’re going to do far more harm than good.

But - speaking of friends - I got to spend this weekend with real friends.

What a treat!

We spent the long weekend at the shack (or as my American friends call it, ‘the cabin’) with my whole family. One grandmother. Two parents. Four married sons and their wives, and 9 children.

Busy but glorious.

As you can see below, it was all a bit hectic for my youngest, Phineas, who decided to tap out during the afternoon book reading.

I trust you have a great week in the classroom.

Happy teaching!