Be part of JetBrains PHPverse 2026 on June 9 – a free online event bringing PHP devs worldwide together.

Eluknow's avatar

Long-time Laravel developers, are you now using AI to code?

Hi all, I've got a question... Now that AI is everywhere, I'm wondering if you guys Long-time Laravel developers, have you changed your habits since the rise of AI?

I'm asking because I am still coding with VS Code like I've always done, and sometimes asking a few things at ChatGPT or equivalent. But I have never installed Claude Code, Cursor, AI agents... And I thought it was "normal" where have been developing this way for so long.

But during the last Laracon, I saw that even Caleb Porzio changed from VS Code to Cursor. And since, I cannot stop thinking about: am I missing something? Should I start working with AI? Is it the new way to code?

That's a real question : I have always thought AI was nice for beginners, to help with basic tasks and learning. But it seems it can do a lot more and some really great developers use it. I am also afraid to use it and "forget" how to code : if AI writes for me, what I'm useful for?

So, do you have any opinion on this?

Thanks,

0 likes
10 replies
Glukinho's avatar

The more AI monkeys are around - the more real specialists are valuable.

2 likes
jlrdw's avatar

So, do you have any opinion on this?

Yes, keep using the keyboard. Only use Ai to help when you are stuck.

Example, I recently got AI to assist in changing some long analytic geometry code to python numpy code.

def calculate_bend_rotation(p_prev, p_curr, p_next, p_after_next):
    v1 = p_curr - p_prev
    v2 = p_next - p_curr
    v3 = p_after_next - p_next

    n1 = np.cross(v1, v2)
    n2 = np.cross(v2, v3)

    dot_product = np.dot(n1, n2)
    magnitudes_product = np.linalg.norm(n1) * np.linalg.norm(n2)

    if magnitudes_product == 0:
        return 0

    angle_rad = np.arccos(np.clip(dot_product / magnitudes_product, -1.0, 1.0))

    direction_indicator = np.dot(v2, np.cross(n1, n2))

    if direction_indicator < 0:
        angle_rad = -angle_rad

    return np.degrees(angle_rad)

I know the math the long regular way but I am not that good yet using the Numpy import in python.

However in a regular crud app in laravel I see no need for AI assistance, laravel is easy enough anyway.

And there's no telling how many business rules would be wrong if done by AI.

Ai doesn't actually code, it compiles other code it scraps to form an answer basically.

Look at it like this:

If humans no longer coded, there would be no new code or ideas of how to do something. Ai is only going to use what exist already. So no new breakthroughs would take place.

Merklin's avatar

I do not use it for coding. Although it can help a lot, each project has its specifics, and each developer has their quirks. Sometimes (quite often) it can lose the context you've provided.

But ... it can help with some menial tasks. Let's say you have a big CSS file you want to split into smaller and more defining/context files, create some documentation or help with repetitive tasks. But even in this case, all jobs are checked by me after.

So yes. As someone has said, I consider it like an assistant to help with some things, but I still prefer to do the main parts by myself.

AhmedHaroon999's avatar

if I do, I know and have grip, otherwise I cannot... AI is posting from around the world so it CAN BE mixture of various approaches which CAN confuse us.

kerelka's avatar

I do it as an assistant for general tasks that require extensive coding.

I only use it in ways that I know what it does and how it works, and I don't create anything I don't know how to do.

jlrdw's avatar

@kerelka

I don't create anything I don't know how to do.

If everyone used it like you, there should be no problem. 😊

I worry about the ones who use it who don't know what they are doing. That could make some very insecure code.

1 like
vinicioswentz's avatar

I'm not a Laravel developer (I'm learning laravel in my free time), but I can give my two cents.

I work with Clojure, and we're free to use the AI tools that the company provides. I was skeptical about them in the beginning, but AI tools are awesome for repetitive work like writing some tests and documenting code. I found that I can explain the type of test I want to an AI tool, and it will write it for me. I then check if it makes sense and write the feature based on the generated test.

It's awesome how much time I'm saving, and the quality of the generated code is fair enough. It's rare that I need to rewrite a generated test.

Another interesting use case I'm testing is using AI to understand features within large codebases. The AI can trace the call chain with incredible speed, summarize the code's key points, and even provide a diagram for better visualization.

These two applications are saving me a significant amount of time, particularly when I have to work in a context I'm not familiar with.

To conclude, no, AI will not make you dumb. It's a tool; play with it and find the best use case for your daily work.

lsvagusa's avatar

Used it extensively during 1 gig where the company explicitly pushed for extreme AI usage as an attempt of battling tight deadlines. In the end it ended up creating more harm than good because of no quality control. So yeah .. after seeing how AI usage works in an actual "crunch time" environment I gave it up for good.

martinbean's avatar

I got passed over for a job because I solved the problem by actually reading the error message and documentation to fix a problem, rather than asking Copilot in VS Code.

After I’d solved the problem, and at the end of the interview, they just kept asking me “why did you not use Copilot?” but in a different way each time, and I got a bit fed up of just answering the same question over and over when they’re ignoring that I still completed the task. So they were clearly after an A.I. jockey.

1 like

Please or to participate in this conversation.