Why Would a Child Learn to Think?

In 1913, Thomas Edison said books would soon be obsolete in schools. Motion pictures would do the job, and the whole system would change within ten years.

It didn’t. Radio was going to turn the world into one big schoolroom. Then came teaching machines, interactive whiteboards and a laptop for every child. Each arrived promising to change education, and each ended up as a tool sitting in a classroom.

I recognise the pattern because I’ve been inside it. We incorporated Zondle in 2012, a platform where teachers set questions inside games. Reading about the answer-marking machines of the 1930s, I realised I’d built a teaching machine with better graphics.

We never thought Zondle would transform education. It was a tool for teachers to pick up when it helped and put down when it didn’t. We feel the same way about oodlü. The technologies that lasted were always the ones that ended up in a teacher’s hands.  They didn’t try to replace the critical human aspect of the teaching process.  The person-to-person exchange that’s lasted the test of time, over the entirety of the human species.

AI is different, though, and that’s why I wanted to write this post.

Everything before it handed something to the child, and the child still had to do the thinking. But AI can do the thinking too. It writes the essay and solves the equation in seconds.

So here’s the question that bites:

If a machine can think for us, do children still need to learn how to think for themselves?

I strongly believe they do, for many reasons.  Not least these:

Firstly, at least for now, you need a deep understanding to use AI well, and to check what it gives you. I build software, and the people who get the best code out of AI are experienced engineers. They know what to ask for. They can read the output and spot the code that works today and falls over at scale, or leaves a security hole. A junior gets something that runs and has no way of knowing whether it’s any good. I suspect the same is true in most professions. AI makes deep knowledge more valuable, because it takes deep knowledge to steer it.

Secondly, understanding is a different thing to thinking, and it gets built through struggle. The research behind our own learning framework calls it productive failure. Give learners a problem they haven't been taught how to solve, let them wrestle with it and get it wrong, and only then teach them the answer. Research on this approach has found they come away with a better grasp of the concepts than learners who were taught first and practised afterwards. The wrestling is what makes the explanation stick.

Put those together, and there’s a problem waiting for us in years to come.

Senior engineers used to be junior engineers. They got there by fixing small bugs, writing the dull parts, getting things wrong and being corrected, year after year. That's the work AI does most easily, so it's the work companies are starting to stop paying people to do. UK graduate vacancies have fallen sharply over the past year. Some of that is down to employer costs rather than AI, but the direction worries me.

Each company’s decision makes sense on its own, but when today’s senior people retire, and the decades of experience needed to replace them never get built because nobody was paid to do the apprenticeship, all that is left is the AI.

School has the same problem, but earlier in the cycle. The junior years at work and the struggle at school are the same slow, effortful stage where expertise starts.

Which brings me to the harder pill to swallow. Human nature.

Thinking takes effort. Most of us take the easy route most of the time, and I include myself.  I call it the “doughnuts and duvets” principle. A child with a deadline and a tool that finishes the work in ten seconds has very little reason to do it slowly themselves.

Young people know that AI could harm their thinking, and use it anyway. Knowing has never been enough to change behaviour.  After all, everyone knows that smoking, drinking and fast food are bad for them, yet these are some of our biggest industries.

If nothing changes, we risk ending up with a generation with all the answers, but no understanding.  In turn, they are less employable for the jobs that are already not there, and within a generation, the human race has deskilled.

I'm not without hope, though. Most of us have seen a child think hard when they wanted to, whether it was over a puzzle, a model, a story or a bike that wouldn't work.

Nobody has to persuade a child to struggle with something they care about. They'll fail thirty times in a row, adjust and go again, because the problem is theirs, the goal matters to them, and failing costs nothing but another go.

I don't know how we recreate that across every subject and every classroom, and I doubt anyone does yet. But a problem the child owns, a goal they want and a failure that's cheap to try again feels like the right place to start looking.

Which brings me back to teachers. The reason to think has always been that someone cares what you think. A teacher asks why you said that, and waits for the answer.

People have taught children face to face for as long as there have been children. Every technology in this post arrived, promised big things and settled into a corner of the room. The teacher stayed.

I think that's because a teacher understands what one particular child needs, on one particular day, and no software can copy that.

It reminds me of seeing a doctor in person compared with over the phone or on a video call. A patient comes in with one problem, and it's only on the way out, when the doctor asks how they're actually doing, that the real one comes out. Remote appointments very often miss that moment. Classrooms have it every day. A child says they've forgotten their homework, and a teacher who knows them hears something else.

I don't think there will ever be a software version of that. It's a human connection, and I'm sure it has saved many lives over the years.

If the machine does the thinking, wanting to think has to come from somewhere else. My bet is that it comes from people.

We'd love to hear your thoughts on this. Find us on the social channels linked at the top of the page.