How to use AI to learn programming faster, not shortcut it
AI can speed up learning to code or quietly stop it. Learn how to use AI as a tutor, with example prompts and how Stuart on start.dev makes the help personal.
AI can make learning to code much faster, or it can make you feel like you are learning while you quietly are not. The difference is not which tool you use. It is whether you let the AI do the thinking for you, or use it to get better at thinking for yourself.
This guide shows how to use an AI assistant as a tutor: how to ask, what to ask for, and the habits that keep you building real skill. It also shows how Stuart, the AI tutor built into start.dev, is designed to make that kind of help personal to you.
The trap: code you cannot explain
The easy way to use AI is to paste in an exercise, get a working answer, and move on. It feels productive. The code runs, the checkmark appears, and you have learned almost nothing. The same problem is waiting for you tomorrow, only now you have less patience for it.
A useful test is this: after you finish, can you explain every line to someone else, and could you rewrite it from scratch tomorrow? If not, the AI did the learning for you.
The rule: try first, then ask
Struggling with a problem for a few minutes before asking for help is what makes the answer stick. Make a real attempt first, even if it is wrong. Then bring that attempt to the AI. You get a more useful answer, because the AI can respond to what you actually wrote, and you get far more from it because you already know what you are missing.
Seven ways to use AI as a tutor
1. Ask for a hint, not the answer
Tell the AI explicitly not to solve it. Specific instructions make a big difference.
I am trying to write a function that returns the largest number in an array. Here is my code and the error I get. Do not give me the solution. Give me one hint about what to check next.
If the hint is not enough, ask for a bigger hint. Work up the ladder instead of jumping to the top.
2. Have it explain errors
Error messages are some of the most useful feedback you get, and some of the most confusing at first. Paste the full message and your code and ask what it means in plain language.
const user = { name: 'Ada' }
console.log(user.address.city)
// TypeError: Cannot read properties of undefined (reading 'city')A good explanation here is that user.address is undefined, and you cannot read city from undefined. Ask the AI to explain the cause, then fix it yourself, for example by checking that user.address exists before reading from it.
3. Explain it back
After you learn a concept, explain it to the AI in your own words and ask it to find what is wrong or missing. This works well because it forces you to recall the idea instead of just recognising it.
Here is my explanation of how closures work. Point out anything incorrect or unclear, and ask me one question to test my understanding.
4. Ask for practice questions
Ask for exercises on exactly the topic you are weak at, and ask it not to include the answers until you have tried them.
Give me five small exercises on JavaScript array methods, from easy to hard. Do not include solutions. I will send my answers afterward for feedback.
5. Ask for a code review
Once your code works, ask the AI how it could be clearer or simpler, and why. This is a great way to learn style and naming. Treat the feedback as suggestions to evaluate, not rules, and ask it to explain any change you do not understand.
6. Ask "why" and "what if"
When something works, ask why. Then change something and predict the result before you run it.
Why does this loop need
letinstead ofvar? What would happen if I usedvar?
Then go and run both versions yourself to check.
7. Compare approaches
Ask for two or three ways to solve the same problem and when you would choose each one. This builds judgment, which is the skill that matters most as tools get better at typing code for you.
Why a tutor that knows your context works better
Every tip above has the same hidden cost when you use a general-purpose chatbot: you have to explain everything first. Which lesson are you on? What is the task? What have you already tried? Which part keeps confusing you? Leave any of that out and you get a generic answer that could have been written for anyone.
Stuart, the coding tutor built into start.dev, removes that setup work. He lives inside the lesson workspace and can see the context of what you are doing: the lesson you are on, the course it belongs to, the challenge you are practising, and the code you have written so far. So when you ask for a hint, he is already looking at your problem.
Stuart can also keep a memory of where you keep getting stuck. If a topic trips you up again and again, he can take that into account the next time you ask for help, so his explanations get more specific to you over time. Everything on the list above, from hints to explaining errors to practice questions, works the same way with Stuart. The difference is that the help is personalised and grounded in your actual course and code, not a generic reply.
A few ways to use that:
- Ask for a hint without pasting anything. Say "Give me a hint, not the solution" while you are in the challenge. Stuart already sees the task and your code.
- Ask about the pattern, not just the bug. "I keep getting stuck on this kind of problem. What should I practise?" works best when the tutor knows your history.
- Check your understanding in place. Explain the lesson back to Stuart in your own words and ask what you missed.
The rule from earlier still applies. Make your own attempt first, and use Stuart to understand why something failed, not just to receive an answer.
Habits that keep you honest
- Always run the code. AI answers can be wrong, sometimes confidently. If you do not run it and understand the result, you cannot know.
- Check anything that matters. AI can invent functions or options that do not exist, or describe older versions of a library. Verify unfamiliar APIs against the official documentation.
- Keep some practice AI-free. Set aside regular sessions where you write code with no assistant. These show you what you can really do and build confidence.
- Type it, do not paste it. Retyping code you are given forces you to read it carefully.
- Write down what you learned. A one-line note per session, in your own words, makes it easier to spot what is not sinking in.
Signs you are using AI well
You are using it well if you can explain your code, if your questions become more specific over time, and if you need the assistant less for the same kinds of problems. You are using it badly if you cannot start a task without it, or if you feel lost when it is not available.
This matters for your career, too. Employers can tell the difference between someone who can direct a tool and someone who understands what it produced. For more on that, read why junior developers can be well positioned for AI and what to do after you vibe-code your first app.
Pair AI with structured practice
AI works best when you already have a path to follow. Without one, it is hard to know what to ask next. A structured course gives you the order of topics and exercises that check your work, and Stuart fills in the gaps, explaining, quizzing, and answering questions along the way, with the lesson and your challenge in view.
If you are starting out, begin with JavaScript Basics (opens in a new tab) and use the habits above whenever you get stuck. Once you are comfortable with JavaScript and want to build with AI yourself, Learn Vercel AI SDK (opens in a new tab) shows you how to generate text, stream responses, and work with structured output in your own code.
