Last week I wrote about the difference between AI augmentation and amplification. Augmentation makes you faster at the same tasks. Amplification makes you capable of things you genuinely couldn't do before.
The responses were interesting. Lots of agreement. Lots of "yes, this is what we should be building toward." And then everyone went back to optimizing their ChatGPT prompts, including me.
Which is fine. Prompts matter. Productivity matters. But here's what's bothering me: we're all so focused on how to use AI that we're missing the bigger question. What are we using it for?
And the answer, whether we admit it or not, is this: we're using AI to make broken systems run faster.
- Education: Schools that were already failing to prepare people for work? Now they're failing faster, doubling down on testing and deploying AI detection tools.
- Work: Managers who were already drowning in impossible expectations? Now they're drowning faster, with AI-powered productivity tracking.
- Access: Access to opportunity that was already restricted to those who could afford it? Now the gap's widening faster, with AI amplification going to whoever can pay for it.
The technology isn't the problem. The systems are.
And until we're willing to name that out loud, all the AI literacy training in the world won't matter.
Why Schools Can't Prepare You for This
Let's start with the obvious: education is designed to sort people, not develop them.
The whole model runs on a simple chain: learn → test → credential → work. You absorb information, prove you can recall it under pressure, get a piece of paper that says you did, and use that paper to access the next level.
It's not a bad system for an industrial economy where the point is to slot people into predefined roles. But it falls apart completely when ChatGPT can ace your exams better than you can.
When the half-life of skills is collapsing, when what you learned two years ago is already outdated, the credential model stops working. You're not preparing people for jobs anymore. You're preparing them for continuous capability development.
Which schools have never been designed to do.
I'm not blaming teachers. Most of them know the system is broken. They're just stuck inside it. In the UK, teachers work an average of 50+ hours a week, but only spend 20% of that time in the classroom. The rest? Lesson planning, marking, admin.
The real question is: if AI can handle the information recall part, what's school actually for?
My hunch? It should be about developing the capability to learn, adapt, and build judgment in uncertainty. Not "here's what everyone needs to know" but "here's how to figure out what you specifically need to develop."
But that requires a complete redesign. Not just new tools in old classrooms. A fundamental rethinking of what education is supposed to produce.
The problem? Most institutions don't want to redesign. Because the current system (for all its failures) serves a purpose. It sorts people. It credentials them. It maintains hierarchies. But there's something deeper going on too.
The people running educational institutions value knowledge retention because that's what they value in themselves. They got where they are by being good at remembering things, at passing tests, at climbing the credential ladder. Teach what you know, right?
And now we're asking them to build a system that says all of that matters less than the ability to learn, adapt, and build judgment in uncertainty. We're asking them to devalue the very thing that made them successful.
Those who run our schools are just as under threat from change as the rest of us. So why would they change?
If everyone could develop capability at their own pace, in their own direction, with AI-powered personalized learning... what happens to the sorting function? What happens to the credential gatekeepers?
The technology exists to build that world. The systems don't want it (and who could blame them).
Why Managers Can't Lead in This
Now let's talk about work.
Every think piece about AI in the workplace says the same thing: "Leaders need to embrace AI and reimagine how work gets done."
Sure. Great. But here's what nobody's saying: we've already made management an impossible job.
Over the last two decades, we've expanded the manager role to include: strategic planning, people development, performance management, compliance oversight, culture building, change management, budget ownership, cross-functional coordination, and about fifteen other things that used to be separate functions.
Most managers are barely keeping up. They're drowning in expanded responsibilities, tighter margins, more metrics, more meetings, more of everything. And now we're asking them to also reimagine the future of work while hitting this month's P&L targets and getting through the performance reviews HR is making them do.
It's not that they don't want to. It's that the system isn't set up for them to succeed.
And people are noticing. Gen Z is 1.7 times more likely than previous generations to turn down leadership opportunities to protect their wellbeing. They're looking at what management has become and saying no thanks.
So they default to what they know: productivity. Output. Control. Because that's measurable. That's defensible. That's what keeps them employed.
And AI fits perfectly into that frame. It makes people faster. It increases output. It helps you do more with less. Which is exactly what leadership wants when they're under pressure to cut costs and prove ROI.
But here's the thing: productivity and capability are different things.
Productivity is about doing more of the same, faster. Capability is about becoming able to do things you couldn't do before. And if your goal is just "do more faster," AI becomes a tool for exploitation, not amplification.
Real amplification (the kind that actually develops human potential) requires slowing down. It requires investing in learning that doesn't have immediate ROI. It requires managers who can recognize and develop capability instead of just tracking output.
But we've built a system where managers don't have time for that. Where quarterly results matter more than long-term development. Where "doing more with less" is the only strategy that gets funded.
So of course AI ends up being used to squeeze more productivity out of people. That's what the system rewards.
The question nobody wants to answer: what would have to break before we could build something different?
Why Access Won't Democratize
This is the part that keeps me up at night.
If AI amplification genuinely works (if it can actually develop human capability at scale), then the people with access will become unreachable by the people without it.
We're not talking about a skills gap. We're talking about a permanent capability chasm.
Think about it: if some people get personalized AI tutoring from childhood, continuous capability development throughout their careers, and amplification tools that make them exponentially more effective... what happens to everyone else?
They don't just fall behind. They become structurally unable to catch up.
This isn't a new problem. It's the same inequality we've always had, just faster and more permanent.
And it's already happening. Right now, there's a private school chain called Alpha School charging $75,000 a year in San Francisco ($65,000 in New York) for AI-powered education. They've replaced teachers with AI tutors, claim their students "learn twice as fast," and rank in the "top 1% nationwide." They're expanding rapidly across the US.
Meanwhile, many public schools are banning AI tools entirely and doubling down on testing.
Same technology. Two completely different worlds, separated by who can afford access.
Because if AI amplification is genuinely transformative, it's only valuable as long as it's scarce. If everyone has it, it's not an advantage anymore. It's just the baseline.
So the people with power right now (the people who can afford access, who control the platforms, who decide who gets amplified) have every incentive to keep it restricted.
This isn't a conspiracy. It's just how power works. When resources get tight, the grip gets tighter. And as economic uncertainty grows, as climate pressure increases, as political instability rises... the people at the top aren't going to voluntarily give up their advantage.
What This Means for the Rest of Us
Here's where I'm supposed to offer solutions. Five steps to fix education. A framework for enlightened leadership. A policy proposal for democratizing access.
I don't have those. (Hell, even I'm charging for access!)
What I have is this: most people already know the systems are broken.
- Teachers know the model doesn't work.
- Managers know they're being asked to do the impossible.
- Workers know the productivity treadmill is unsustainable.
What we lack isn't permission to say it out loud. What we lack is control. Agency. We live in a world where decisions are made by people with power, with little consideration for the ramifications on the rest of us.
And maybe that's the work right now. Not optimizing AI adoption. Not getting better at prompts. Not pretending that individual capability development will save us from systemic failure.
Maybe the work is finding each other. Building community (online and offline) with people who see what you see. Crafting our identities together. Finding refuge in the knowledge that we're not alone in this.
Because you can't redesign something if you're still pretending it works. But you also can't redesign it alone.
I don't know what the rebuilt system looks like. I don't know if we get there through gradual reform or through something breaking first. I don't know if the people with power will choose to share it, or if it'll have to be taken.
But I know this: the conversation we're having about AI right now (about productivity, about augmentation, about "learning to use the tools") is a distraction from the conversation we need to be having.
Which is: what are we actually building? Who is it for? And what are we willing to let break in order to build something better?
Those are harder questions than "how do I write a better prompt."
But they're the ones that matter.