AI Is Making Students Worse at Learning — Here's Why That Matters

I want to say something uncomfortable:

AI is making it harder for students to learn.

Not in the obvious way — not because the tools are confusing or broken. But because they work too well. Every time a student gets stuck, they open ChatGPT, paste the error, get the fix, and move on. The problem is solved. The learning didn't happen.

The Problem: The Struggle Was the Point

When I was learning to code, getting stuck was the curriculum. You would stare at a bug for two hours, try six different things, fail at all of them, and then — maybe — find the answer. That two hours of failure was where the actual learning lived. You built mental models. You learned why things break. You remembered.

AI removes that entire process. The student goes from stuck → solved in 10 seconds. The code works. But ask them the next day why the fix worked, and they cannot tell you.

What we are seeing is not accelerated learning. It is answer acquisition. And those are not the same thing.

Cognitive science has known for decades that the "testing effect" — struggling to retrieve information from memory — is what builds long-term retention. AI bypasses that retrieval step entirely. The student gets the answer, feels productive, and learns nothing durable. Speed without retention is just busy work.

The Historical Pattern We Are Ignoring

This is not the first time technology has been sold to us as a productivity tool while quietly making us worse at something important.

In the late 2000s, social media platforms were marketed exactly the same way AI is marketed today:

• "Connect with colleagues and share ideas professionally."
• "A powerful networking tool for your career."
• "Stay updated with industry thought leaders."

Fast forward fifteen years and every phone ships with a "Screen Time" or "Digital Wellbeing" feature designed to limit the exact apps that were supposed to make us productive. We had to build cages for the tools we invited into our lives.

AI is following the same trajectory. Right now, every company is marketing their AI assistant as a productivity multiplier. But for students — for learners — that framing is wrong.

When you give a student an AI that writes their code, their essays, and their problem solutions, you are not making them productive. You are making them dependent.

The Environmental Cost Nobody Talks About

There is another layer to this that gets overlooked. Every AI query consumes energy and water for data center cooling. Estimates vary widely — older analyses put ChatGPT at 5-10x the energy of a Google search, while newer model optimizations have narrowed that gap significantly. But even on the conservative end, a student asking AI what they could look up in documentation is still a net increase in compute waste.

Most student queries — "what does this error mean," "explain this concept," "write a function that does X" — could be answered by a web search, a documentation page, or a textbook. The AI is not unlocking new knowledge. It is replacing the act of looking something up. And that trade-off is rarely worth the environmental cost.

What Should Governments Have Done?

I believe governments should have stepped in much earlier on AI regulation — not to stop innovation, but to control distribution, especially in education. We did not wait fifteen years to regulate social media's impact on teenagers. But with AI, we are repeating the same mistake. The horse is already out of the barn.

Some countries are starting to act. Italy temporarily banned ChatGPT in 2023 over privacy concerns. The EU AI Act is trying to create a regulatory framework. But these efforts are reactive and slow. The damage to how a generation approaches learning is already happening.

What Students Can Do Now

Since regulation is not coming fast enough, the responsibility falls on individual students. And I know that is an unfair burden to place on someone who is just trying to get through school. But here is what I would recommend:

1. Treat AI like social media, not a calculator

Put ChatGPT and Copilot in the same category as Instagram and TikTok — tools that can consume your time and replace your thinking. Use your phone's app limits if you have to. Block these tools during study hours.

2. Implement a "try-first" rule

Before you ask AI for help, spend at least 15-20 minutes trying to solve the problem yourself. Search the web. Read documentation. Try a wrong approach and see why it fails. Only after that should you bring in the AI.

3. Use AI as a tutor, not a solution generator

Instead of "write a function that sorts this array," try "explain the different sorting algorithms and when each one is appropriate." One gives you answers. The other helps you understand.

4. Be honest with yourself about what you actually know

At the end of the week, ask yourself: can I explain the concepts I "learned" without AI assistance? If the answer is no, you did not learn them. You just collected answers.

The Bigger Picture

AI is an incredible technology. It has genuine use cases in research, healthcare, accessibility, and automation of genuinely tedious work. But not everything that is powerful needs to be everywhere.

For students, the most valuable skill is not the ability to get the right answer quickly. It is the ability to solve problems they have never seen before, with tools they understand deeply. AI shortcuts that process at exactly the wrong moment — when the foundation is being built.

We need to stop pretending that giving every student an AI assistant is an unqualified good. The tools are here to stay. But how we use them — and how we teach students to use them — will determine whether this generation of learners emerges stronger or more dependent.

The choice is ours. But we need to start talking about it honestly.