We recently released more than ten thousand illustrations and worksheets for children, free of charge. Alongside them are 340 four-panel comics on digital citizenship, each with a narrated video.
The truth is, most of those materials were never meant to be published.
They were made because we needed them in-house, to keep producing the comics at a consistent quality. The same characters. The same expressions. The same backgrounds, the same objects. Asking an AI from scratch every time, you will never reach 340 episodes. The library itself was the machinery that kept the comics coming.
We opened it up because several people working in education wrote to ask whether they could use the materials. Better in their hands than kept as an internal tool. That was the judgement we made.
You might look at it and think, "with AI, anyone could make this given enough time." People who use AI daily, on the other hand, tend to say, "that must have been an enormous amount of work."
The same thing, seen so differently. Where that gap comes from is what I want to write about today.
Chance gets you one picture
Making a single picture with AI is something anyone can do now. With a little luck, the first attempt is startlingly good. Feeling "I could do this too" is a perfectly reasonable response. That is using AI by chance.
Three hundred and forty comics is another matter. Chance will not carry you a single step.
A four-panel comic is not one picture. The same person has to keep the same face across four panels, the story has to connect, and something has to remain with the reader at the end — all at once. Pulling the lever until a good picture appears means episode one and episode three hundred belong to different works.
So we built the materials first. Fix the characters, lay out the expressions, prepare the backgrounds and the props. Only then can you keep producing at one level. Ten thousand is not a count of finished works. It is the size of the foundation we needed in order to work steadily.
And here is the part that matters most. Producing a great deal at once is far easier than producing at the same level over time.
The first ten can be made on willpower. The problem is the hundredth. The three-hundredth. Quality not slipping there, characters not drifting, no episode where somebody cut a corner. This is not a question of perseverance. Keeping going takes a system, not stamina.
What you need is design, not effort
So what does it take?
What to do, in what order, and how much to hand to the machine. Where a person decides, and what gets checked automatically. What to keep as a pattern so that nothing is built twice. You settle these first, and turn them into a program.
Producing the picture is a small part of the whole. Most of the time goes into how everything around it is assembled. And deciding what you are aiming for, and whether what came out is good enough — that part a person has to carry. Only you know what you were trying to make.
It is true that AI made it possible for anyone to produce something. It is equally true that, because anyone can, the difference now lies in how you produce it.
The question is whether you can do it again
This is not confined to making teaching materials.
Anyone can now turn out something respectable. Writing, pictures, documents — as one-offs, they no longer separate people the way they used to.
To be clear, this is not to say that what someone made has no worth. It is that a single piece of work no longer shows you much about the person who made it.
What is asked instead is whether it can be produced again, and again. People who can produce one good document are everywhere. People who can produce them repeatedly are not. What separates them is not talent but whether they hold a system.
The gap is invisible in calm conditions. When one will do, when the deadline is generous, when nobody asks for changes, chance and design wear the same face. The difference appears when volume is required, when a deadline bites, and when someone says "change this part." Those relying on chance stop there. All they can do is wait for the next lucky pull.
Thinking by design comes only from repetition
Here is the awkward part.
Do something once and you cannot see what made it work. Do it many times, compare the days it went well with the days it did not, and "this is how you get there" begins to show itself. That is what it means to say experience is required.
What that needs is a place where failing is fine, and enough time to go round again.
Using AI by chance, or by design. The difference does not arrive in a single sitting. Only the children who get to try many times make it their own, little by little.
Digital Kodomo BASE runs a free learning space, After-School Lab, every Wednesday, with one computer per child. There is also Furatto Lab, for children who find school hard and for their parents. For children with no computer at home, we give and lend them.
What we want to hand over is not a method for making things well. It is the time in which a child can say, after it did not work, "let me change this bit and try again."
We want to be the place a child still wants to come back to, even on the walk home from a day when nothing came out right.