Is using AI to build projects without knowing web dev a bad idea?

Posted by Character_Cold4105@reddit | learnprogramming | View on Reddit | 11 comments

I’m building a project rn, but I don’t really know web development. I’m using AI a lot to help generate & explain the code, & then I try to understand. I wanted to ask if is this a bad habit in the long run, or can this actually be a valid way to learn? My concern is that I might become too dependent on AI and struggle later in internships/jobs when I have to build things on my own. But at the same time, using AI is helping me build projects.

What’s your honest take on learning/building this way? Have any of you learned like this successfully?

[-]

mandzeete@reddit

Okay, you are coming with your project. I will be asking from you:

1)Why did you build it? What purpose does it serve? What can you answer to these questions? Do not build a project just for the purpose of building projects or for the purpose of having projects. It is quality over quantity. If you can't talk about the stuff you built then as well these things do not exist in my eyes.

2)Can you make this or that change? A person who built his project is able to make changes. Fine, perhaps if it is an old project the guy must take some minutes to read the code, but after that he'll remember stuff and is able to make the changes. Will you be able to make changes in a "project" that you did not build?

3)Can you tell me why did you do this or that? A person who built the project is able to explain why he did the thing as he did. In your case, you did not write the code but it was the AI. Can you explain the (lack of) reasoning behind the choices that AI did when it wrote the project the way it wrote?

4)Can you ensure the code is safe and scalable? Fine, a beginner is often making as bad code as the AI does. But later on, at work, do not expect your code changes to be approved when (not) you generate a code that introduces different vulnerabilities or is a whole lot of nonsense hack.

5)Can you guarantee that nothing will break in production when you submit your changes? That goes when you are working already. You can't say "But the AI did it." No. It was you who did it not the AI. When something breaks, when the client or end user suffers some damage or data loss or such, then it is you who caused it not the AI tool you used.

6)Can you build it without an AI? Not every client permits LLM-based tool usage. For whichever reason. It can be geopolitics related (our company has banned Deepseek, for example). It can be about client wanting to protect his business secrets. It can be very sensitive project. Or such. What will you do then?

You can use the AI when you are able to explain what it does and why it does. When you are able to push back the changes it suggests and the nonsense it tries to do.

[-]

MeLittleThing@reddit

What's the best way to learn cooking? Cooking or ordering food?

[-]

Character_Cold4105@reddit (OP)

It's like I am learning cooking with a chef beside me giving instructions

[-]

AlwaysHopelesslyLost@reddit

LLMs are not chefs. They cannot think, reason, or be creative.

[-]

pointlesslyDisagrees@reddit

You can't define any of those terms

[-]

taedrin@reddit

If you are using AI to generate code, then the you are learning to cook by watching a chef cook. Which is better than nothing, but it's not like you are going to become a competent chef if the only thing you do is watch "Binging with Babish" everyday.

Sooner or later you need to take off the training wheels and learn how to do stuff without AI assistance, or else you aren't going to develop the necessary skills to be able to direct the AI instead of be directed by the AI.

[-]