Can we stop recommending AI tools to people who can barely create a variable?

At least, can we stop suggesting using them wrong?

There’s a right and wrong way to use AI for learning, but I feel like no one is talking about it. There’s a difference between asking ChatGPT for help to understand something and using it to solve your homework. In this article…

1. The right and wrong way to use AI tools
2. How to use AI if you’re a coding student
3. Why AI should not be used by students (and maybe not even by juniors)

Let’s get something out of the way: Programming is not for everyone.

Similar to other crafts and hobbies, if you don’t like it, it will get old. Very quickly. And you will go crazy.

If you don’t like programming, simply, don’t pursue a career in programming. Don’t do it for the money. You will never see a fat paycheck if you don’t like programming because if you don’t enjoy it, you will always suck at it.

Now that that’s out of the way, let’s talk to people who do like programming. If you're trying to learn, be careful about how you use AI. At your level, AI should not be used to delegate. It should be used, if at all, to better understand things that might be unclear. If you’re in school, you shouldn’t need ChatGPT. You should take advantage of the living, breathing teachers and TAs available to you—that you’re likely already paying. Relying on ChatGPT when you have access to real-life support is just silly, honestly.

Now, if you’re not in school or don’t have access to a mentor, AI tools can be useful, but you need to be cautious: AI tools can really, really hinder your understanding and learning curve.

Programming isn’t just muscle memory or repeating the same four tasks in the same exact way for the rest of your career. It’s developing mental schemes and logical patterns to solve complex, challenging problems. How do you develop this skill?

By making mistakes. By banging your head on the keyboard for three hours just to realize you wrote j++ instead of i++. That’s just an unavoidable part of learning anything. Delegating basic tasks like “write a loop to sum all the elements in a given array” because you think they’re beneath you is just lazy.

You will never learn anything by not making mistakes and by not putting in the crazy effort it takes to learn something as nuanced and complex as programming.

Do you really need to use AI to learn how to code? No. Everyone who learned before the LLM era did just fine. Sometimes, better. And I know I sound like a boomer saying “I got beat up as a kid and I turned out fine”, but this is just the truth. You do not need AI to learn how to code. Period. There’s endless amounts of information and documentation out there you can read.

But AI can be useful in everyone’s life, or it wouldn’t even exist. Most people just need to be educated on how to use it properly.

How to use AI tools if you’re a coding student:

Let’s start with the basics. ChatGPT shouldn’t write your code. EVER.

ChatGPT is the equivalent of hitting the suggestion button on your phone until you get a sentence together. It’s not intelligent; it doesn’t think. It’s mix-and-matching code from the internet and copy-pasting it.

And yes, that might include GitHub repos from other students—repositories filled with errors, bugs, and horrible coding practices. ChatGPT’s code is trash, y’all.

ChatGPT is good at summarizing and grouping information about a topic (not always, since it hallucinates a lot), but you should never use it to write your code or do your homework for you. It’s counterproductive for your learning and introduces the risk of bugs.

If you’re a student doing homework, listen up:

I’ve been a teacher my whole career. It’s my main gig. And there’s one foolproof way to tell when someone has copied their homework and put zero effort into it. If we just had a class on for-loops, why is your homework done using ES6 methods? Why is your homework done in Java instead of JavaScript? (Yes, I’ve had that happen.)

Y’all don’t fool anyone. We know. 👀

Now, let’s discuss how to use AI correctly when learning something new. Here’s what I do every time I’m struggling with a new concept:

• Don’t ask for a solution; ask for an explanation. Even if ChatGPT gives you the solution anyway, ignore it. ChatGPT often gives a brief explanation of the problem before writing code for you. Read the explanation. Think with your own brain. What does the explanation entail?
• Ask additional questions! Let’s keep the previous example. If you’re wondering, “What is an iteration step?” ask that! Curiosity is everything when it comes to learning! You should be curious about the craft you’re learning. If you’re not, it’s likely you don’t enjoy it as much as you say. Don’t just settle for “ChatGPT said so, so it must be right.” Always ask yourself more questions.
• Never use AI (or not AI) autocomplete tools like Copilot or Tabnine. No. Just… no. If you’re learning how to write an HTTP request using fetch, you should not be using Copilot to autocomplete it for you. Why? Because to fully understand how fetch works, you need to at least be able to write it yourself. I’m not asking for much—just type it out. It won’t hurt your fingers, I promise. Also, during your job search you might be asked to do some whiteboard coding, and there’s no Tabnine to save you then.
• Always make sure you understand the explanation provided. If it doesn’t make sense, ask for clarification. ChatGPT isn’t going to get bored of you asking over and over. If the logic doesn’t sit right with you, the code will be trash or wrong. Without understanding the logic, you won’t be able to write good code. So, ensure you grasp the logic and put in the effort to figure it out. Work those brain muscles!

ChatGPT doesn’t just mess with your technical knowledge and hinder your learning.

ChatGPT is for your logical skills the same as TikToks are for your your attention span. It f*cks them up.

AI can diminish your ability to think critically, problem-solve, and build resilience in learning. By relying on AI too early, you miss out on the grit and problem-solving process that defines coding. It’s the same reason you don’t want to skip learning how to float when entering a swim competition. The basics and the foundations are essential for building long-term expertise. You need to struggle. YOU MUST STRUGGLE. It’s supposed to be hard. It’s supposed to be riddled with mistakes and sleepless nights. That’s how you get better.