Use AI to Learn, Not to Skip the Work
4/8/2026
One of the biggest mistakes I saw as a coding bootcamp instructor, and still see now, is simple.
Students copy the answer without reading it, trust that it’s correct, and move on.
The problem is they finish the assignment but learn nothing.
And now that AI has gotten so good, this is easier than ever. I’ve seen students hand an entire lab to AI and get back something that works.
If your goal is just to finish, that’s great. If your goal is to actually become a software engineer, it’s a trap.
When I was teaching in a coding bootcamp, the students who improved the fastest weren’t the ones who used AI the most. They were the ones who used it differently.
They treated it like a companion, not a replacement.
They still wrote the code. But they used the AI to do things like help them pseudo code, to organize their thoughts, and to critique what they wrote.
That difference matters more than anything else.
There are two very different ways to use AI when you’re learning.
You can use it to do the work for you.
Or you can use it to help you understand the work.
The first one feels faster. The second one actually makes you better.
One of the most effective things you can do is have AI critique your code instead of writing it.
Write your solution first. Even if it’s messy. Even if you’re not sure it’s correct.
Then ask AI to review it.
Ask questions like:
- What could I improve here?
- Is there anything unclear or confusing?
- How could this be simpler?
This forces you to stay engaged while still getting feedback.
You are doing the reps. The AI is just coaching.
Another important skill is learning how to calibrate the AI to your level.
AI will often give you the “best” answer, but the best answer is not always the most helpful when you’re learning. It might introduce patterns you don’t understand yet or optimize things in ways that make the code harder to follow.
So you need to guide it.
Tell it things like:
- Explain this in simple terms
- Prefer clarity over efficiency
- Don’t introduce advanced patterns
- Keep the solution beginner-friendly
This makes a huge difference. You’re not just asking for answers. You’re shaping how those answers are taught to you.
Another approach I use a lot is having AI outline the logic instead of writing the code.
Instead of saying “solve this,” ask it to break the problem down into steps.
Treat it like a blueprint or even like Mad Libs.
For example, you might ask:
- Outline the steps to solve this problem
- Do not write the code
And get something like:
- Step 1: validate the input
- Step 2: loop through the data
- Step 3: store the result
- Step 4: return the final value
Then you fill in the code yourself.
This keeps you thinking while still giving you direction.
You can also get a lot of value just by asking better questions.
Instead of asking:
“Fix this”
Ask:
- Why is this wrong?
- What is this code actually doing?
- What problem does this pattern solve?
That shift from what to why is where a lot of learning happens.
AI is only going to get better from here.
It will get faster, more accurate, and more capable of doing entire projects on its own.
That means the temptation to let it do everything will only increase.
But the rule stays the same.
If you let it do the thinking for you, you won’t improve.
If you use it to support your thinking, you will learn faster than ever.