"What are you doing?"
"PlantSnap. You photograph plants and it tells you what they are. I'm trying to get all the ones in our neighborhood."
I watched him crouch down, angling his phone to capture a small purple flower by the fence. His face had this focused intensity it has when he is deeply concentrating, but that you seldom see during homework.
This is the same kid who groans dramatically whenever his biology textbook appears. Who comes home and announces, with the theatrical certainty of a nine-year-old, "I'm allergic to biology class. Seriously, I think I'm allergic."
But give him a game where he's hunting plants like Pokémon? He's unstoppable.
That morning, watching him photograph Vicia cracca without knowing that's what he was learning, I thought: What if all of school could work like this?
When the system fits some brains but not others
My friend called me the other day. We talk about a lot of things, but lately often about her daughter being frustrated at school. Over and over again.
"She says she's stupid because reading takes her so long," my friend said, exhausted. "But she's not stupid. She's diagnosed as dyslexic."
I've watched this girl talk about fairytales and movies with such insight it gives me chills. She understands metaphor, symbolism, character development—everything a literature teacher dreams of hearing.
She just can't efficiently decode text on a page. Reading isn't hard for her because she lacks intelligence. It's hard because her brain processes written symbols differently.
But now my friend phoned me for a piece of news. Her daughter is now allowed to use audiobooks for school assignments. Voice notes instead of written essays. And today she told her mom something that made me want to cry: "I think I might actually be good at literature."
She was always good at it. The delivery method was the problem, not her brain.
Or take my colleague's son. Brilliant kid. Actually brilliant—he thinks in three dimensions like an architect. But he fails math tests consistently.
Why? Because math tests are abstract symbols on paper, often timed. His brain works spatially—through building, through seeing how things fit together, through his hands.
Give him a design challenge? He intuitively grasps structural engineering principles that take others hours to understand. Force him to solve equations under time pressure? He freezes completely.
Same mathematical concepts. Entirely different access points.
The pattern I keep seeing
My son hunting plants with his phone. My friend's daughter finally understanding she's good at literature. My colleague's son building structures that demonstrate complex geometry.
Every single person I know learns differently.
And yet our school system picks one approach—lecture and textbook—and hopes it works for enough students.
It doesn't.
What technology might actually change
I saw a presentation last week where someone demonstrated Khan Academy's Khanmigo. It's an AI tutor, but not the kind that just gives you answers.
A student gets stuck on a problem. Instead of solving it for them, Khanmigo asks: "What have you tried so far?" "What do you know about this type of question?" "What happens if you approach it this way?"
It guides. It adjusts difficulty based on how the student responds. It never gets frustrated or impatient. And I know this approach is used elsewhere too. For example Harward University’s CS50 course has also made their own chatbot to help students by guiding them, not giving the direct answer as a code. But if we concentrate now to the Khanmigo, their statistics are striking: students using these adaptive systems show 54% higher test scores. At one university, trigonometry pass rates jumped from 76% to 94%.
But honestly? The numbers aren't what excites me most.
What excites me is the possibility—finally—of matching content delivery to how each brain actually works.
Imagine my son's entire biology curriculum delivered through that plant app. Same scientific content, completely different engagement. He'd be outside every day, hunting species, earning points, competing with friends. Learning without the resistance that comes from forcing his active brain to sit still with a textbook.
Or my friend's daughter accessing every school text as an audiobook with one click. No more fighting her dyslexic brain just to get to the ideas underneath the words.
My colleague's son learning geometry through building challenges and spatial puzzles instead of timed worksheets. Finally being able to show what he actually understands.
AI can present the same concept through radically different formats:
- Video demonstrations for visual thinkers
- Audio explanations for auditory processors
- Interactive simulations for hands-on learners
- Games for kids who engage through competition
- Text for the ones who genuinely love reading
This isn't about "learning styles"—that theory's been thoroughly debunked. This is about the reality that different brains engage differently with different approaches. That's not pseudoscience. That's just honest observation.
The part that worries me
But I have concerns too. Real ones.
What if we build amazing AI education systems and they only work for English speakers? For kids in wealthy schools with good technology infrastructure and the best tech-positive teachers?
From my Finnish perspective as a minority within another small language area I also worry about the user experience in my dear mother tongue Swedish. What if they only understand the Swedish spoken in Stockholm and not the Finland Swedish we (and Moomin) actually speak here? Or what if they completely erase Sami languages from existence?
And here's the scariest one: What if these systems make learning so frictionless that kids never develop resilience? Never learn to struggle productively with hard problems?
Research is already showing that some students using AI get better grades while developing weaker thinking skills. They can produce correct answers, but don't actually understand the concepts. They know how to prompt an AI but not how to reason independently.
That's not education. That's automation of performance without learning.
We need AI that builds independence, not dependence. That guides students toward understanding rather than just giving them answers to copy.
And we need to face the linguistic justice question directly: 90% of AI training data is in English. When systems do support other languages, they give you the standardized version. Rikssvenska from Stockholm, not the Swedish we speak here. Standard Helsinki Finnish, not the linguistic diversity that actually exists in this country.
We're building transformative technology while leaving most of the world behind.
What I want for my own kids
It's Friday evening now. The kids are asleep after a long week.
My son still hates homework when it involves reading long chapters and memorizing lists from books. But he's learned every plant in our yard and half the neighbourhood.
My daughter is eighteen now, in her final year before university. She's had her own battles with a school system designed for most students but not quite for her. She's learned to "hack her own learning"—figured out workarounds, found what actually works for her brain even when it doesn't match the official method.
But she shouldn't have had to fight that hard.
Here's what I hope for—for them, for every kid:
I want my children (and every other child too) to finish school still curious. Still believing they can learn new things. Still thinking of themselves as capable.
I want them to have access to tools that adapt to their brains instead of forcing their brains to adapt to tools. Tools that work not only in English but in Finnish and their mother tongue Swedish—our Swedish, an official langue in this country, with our expressions and our rhythm. Tools that recognize different ways of learning as equally legitimate.
I want my son to learn biology from his teacher, but also through games and outdoor exploration if that works for him. My daughter using her own hard-won methods. Every kid through whatever pathway actually makes sense for their brain.
Not to forget to free the teachers from administrative burden to do the work they signed up for: teaching.
The choice we're making right now
The technology exists. That's not the question anymore.
The question is: Will we build it equitably? Will we prioritize linguistic diversity? Will we design for independence rather than dependence? Will we keep humans at the center?
Or will we outsource our own thinking to the AI or will we just automate the same old sorting mechanisms that already fail so many kids?
Yesterday, my son and I walked to the bus stop. He suddenly stopped at a plant growing through a crack in the sidewalk.
"Kolla mamma, mjölke," he said proudly, pointing. Rosebay willowherb, apparently.
I wouldn't have known that name—in any language. He learned it from a game on his phone.
He still doesn't love homework. He still resists assignments required by him. But he's learning anyway—through play, through curiosity, through an approach that actually fits how his brain works. As I’ve been writing this he was singing a song he heard at an online footboll game, searched for from YouTube and now memorised the lyrics—in Spanish.
That's not lowering standards. That's finally matching teaching to learning.
That's the future I want us to build. Not someday. Now.
Because every week we wait is another week of kids coming home thinking they're "just not smart enough" when really, they just learn in a different way.
And we finally have the technology to do something about that. And to be honest, our network at ÄCID is at the moment working with something to make learning better and easier for many Finnish students ;-)
JeS
P.S. If you want to read a super long, but excellent history of education through the Western history: do not miss Peki’s marvellous blog post From Athens to Algorithms: How We Lost Personalized Learning.


