Life is changing fast enough these days that it's hard not to notice. And it raises an obvious question: what mark will this moment leave on the kids and teenagers growing up inside it?
We've actually been here before, in a smaller way. Before AI became the headline story, the flood of new media and digital content promised us something specific: a "digital democracy," where more connection meant more understanding between more people. What actually arrived was different — a flood of misinformation, confirmation bias, and filter bubbles. We got more connected and, paradoxically, more isolated inside our own separate rooms, all at once.
AI pushes that same trend a step further. Convincing fake video is now something almost anyone can produce. Confirmation bias and filter bubbles are about to get more sophisticated, not less. Conversations with AI are starting to run longer than conversations with other people, and an AI's answer is starting to feel more trustworthy than an expert's diagnosis. Unless you personally know more than the AI does about a given subject, it may soon feel natural to just hand the judgment itself over to it.
We've solved this exact shape of problem before
As transportation got easier and industrial society matured, human physical activity declined. But we didn't just let that gap sit empty — we invented exercise. Deliberately. Today, working out is its own industry, and a toned, trained body has become a cultural marker of a healthy life.
Education hasn't made that same leap. Our education systems are still anchored to an industrial-era model — structured knowledge transfer, and an even narrower obsession with college entrance exams. Local elections for education superintendents happen, but the choice tends to stay trapped in political alignment rather than a real debate about the core problems in how kids are actually being taught.
What AI takes, and what has to replace it
AI is overwhelmingly better than any human at accumulating and retaining knowledge. That doesn't mean personal knowledge stops mattering. It means knowledge alone is no longer enough to be anyone's core competitive edge.
"Just as exercise filled the space left by declining physical activity, the space left as we hand our judgment over to AI will be filled by deliberate training in how to think." I heard that line at a talk once, and I haven't stopped agreeing with it since.
I believe AI-era education isn't about loading in more knowledge. It's about helping people practice asking their own questions, forming their own judgments, and testing those judgments against other people's — building what amounts to a "thinking muscle."
Knowledge is a search away. You can just ask AI. But deciding what to search, or what to ask in the first place — that's not something a search engine or an AI resolves for you. And judging what you get back isn't the AI's job either.
The strength to think isn't something you can search for or ask for. Can we find a genuine daily practice for it — the kind that, like a muscle, only grows through regular, repeated use?
Old models for a new problem
Some regional education offices in Korea have recently introduced discussion-based classes specifically to build students' literacy and communication skills. But many of these classroom formats were designed a long time ago, and there's real skepticism about whether they're actually effective for how students learn today.
This is the gap the Agreement Rate Game was built to close. It's structured so that predicting other people's thinking accurately — not just having an opinion of your own — is what earns a higher score. The mechanic originated in interactive broadcast technology built for audiences of hundreds of thousands, and it took its current shape once it was grafted onto a children's program that needed a fairer way to run a mass, real-time voting game.
Because the group answering is different every single time, even AI struggles to predict the exact agreement rate in advance. That unpredictability is, in a strange way, the whole point: it's what forces genuine engagement with what other people in the room actually think, rather than a guess anchored to what you assume they'll think.
An early, honest verdict
Does the Agreement Rate Game actually build depth of thought, and a stronger habit of considering other people's positions? Mind Catch Class — the classroom format built on this same mechanic, where predicting others' thinking accurately is what wins — is now running as a pilot in a handful of schools, and the early response has been genuinely positive.
That said, there's still real distance between a successful pilot and something that scales across a national education system. The gym for thinking still has to be built one classroom at a time.