The Problem with AI and Our Learning

What exactly is the problem with AI and our learning? The question seems simple, but almost every answer I hear falls into one of two opposite camps. On one side, AI is the revolution that will democratize knowledge. On the other, it is the ruin that will atrophy our ability to think. Neither camp convinces me. And the reason is that both are answering the wrong question.

The question isn't whether AI is good or bad for learning. The question is: what does AI solve, and what does it make easier to avoid? Because every tool that solves one part of the work also makes the other part more optional. And what becomes optional, sooner or later, becomes rare.

The wrong question

The wrong question is the one that asks for a verdict. Is AI good or bad? Should we use it or ban it? These questions feel urgent, but they don't help us decide anything concrete. They treat AI as a single thing, with a single effect, when it's actually a set of tools that touch different parts of learning in different ways.

The right question is more uncomfortable. It asks us to look at what we actually do when we learn, and then to ask which of those things AI can do for us — and which of those things we are now free to skip. That second part is where the trouble lives. Because freedom to skip is not the same as wisdom to choose.

What AI solves

AI solves the easy part. Finding information, summarizing a text, translating a sentence, organizing an argument, recalling a definition, building an example. This part was always the visible part of learning — the part that can be shown, measured, graded. And AI does it better and faster than we do.

This is not small. For someone who never had access to a tutor, a translator, a reviewer, AI is a door that opened. I don't want to minimize that. But the easy part was never the problem of learning. It was always the part that could be outsourced without great loss. What was lost when we outsourced the easy part was little — memory, speed, repetition. Things that can be recovered.

What AI doesn't solve

AI doesn't solve the hard part. The part where you decide what is worth learning. The part where you sustain attention when the subject isn't interesting. The part where you get it wrong and sit with the error. The part where you build your own opinion after reading five sources that disagree with each other. The part where you change your mind. The part where you learn something that serves no immediate purpose.

This part was never visible. It can't be measured easily. It can't be shown on a test. It can't be summarized in a certificate. It's the part that happens in silence, between one reading and another, on a walk, in a conversation that wasn't planned. It's the part that builds what we call judgment — the ability to decide without having to look it up.

What becomes optional

When AI solves the easy part, the hard part becomes optional. And what becomes optional, sooner or later, becomes rare. Not because people are lazy. Because the easy part was always tied to the hard part by an invisible thread. Whoever summarized a text learned to summarize. Whoever translated a sentence learned to translate. Whoever organized an argument learned to argue.

When AI does the summary, the translation, and the organization, the thread breaks. The hard part still exists, but it is no longer necessary to reach the visible result. And the visible result is what most of us chase. Not out of bad faith. Because it's what can be shown.

This is the quiet shift nobody announces. The work doesn't disappear. It just stops being required. And what stops being required, in a culture that rewards speed, tends to stop being done. The hard part doesn't vanish overnight. It fades. It becomes the thing only a few people still do — and then the thing almost no one remembers how to do.

A question I carry

I carry a question with me since I started using AI in my own learning: what am I failing to become when I outsource this? It isn't a question against the tool. It's a question about what the tool makes optional in me.

Sometimes the answer is: nothing important. The summary I asked for wasn't going to teach me anything I needed. The translation saved me time on something that didn't require my attention. And that's fine. Not every task needs to shape me.

But sometimes the answer is harder. Sometimes I realize I asked AI to do the very thing that would have built the judgment I now lack. I asked it to organize the argument I couldn't organize. To find the flaw I couldn't find. To explain the idea I couldn't explain. And in asking, I skipped the exact exercise that would have made me more capable.

That is the problem I'm trying to name. Not that AI is bad. But that AI is good enough to make the hard part optional — and the hard part is where we become who we are. The tool doesn't decide for us. It only makes the decision easier to avoid. And avoiding a decision, over time, is a decision too.