AI is a hot-button issue. Honestly, I might even call it a molten lava issue.
And I’m certainly someone who sees both sides of the issue.
On the one hand, there are many problems with AI, from environmental impacts to the proliferation of misinformation—whether deliberate or due to hallucinations—to copyright infringement and more.
On the other hand, at least for programming and development, it can help speed up dev work, especially the menial tasks.
I spent the last year trying out a few different AI tools in my development workflow. Here’s what I learned and how I’m using it now in my programming and development.
What I’ve Learned
Let’s start with what I’ve learned using AI in my development workflow.
AI Will Not Take My Job
AI is helpful, but it’s not a replacement for real developers.
This is definitely the angle that concerns most people in my profession, but I don’t think we have a lot to worry about right now for established developers.
It is a tool, and it has to be handled properly for it to work. You can not just say, “Create a website or an app,” and have it spit out something perfect.
You need someone who knows what they are doing to write effective prompts that get the results you want, and to review the AI-generated code and spot errors.
There will always need to be a human element in web development, design, and programming.
I’m Glad I Learned Web Development Deeply Before AI
On another note, I’m glad I learned web development and grew as a web developer before AI came along.
I have the hands-on experience necessary to really know what I’m doing. I’ve written a lot of code. I’ve had to search Google for specific problems I’ve run into, hoping someone on Stack Overflow has encountered the same issue, found the answer, and shared it. I’ve had to think about the structure of my apps. And I’ve had to search through documentation.
I have the experience to know what I’m doing. I’m not reliant on the AI to know everything.
And that makes all the difference in creating things that actually work.
You Have to be Specific
As I mentioned before, you can not give the AI generic “create an app” or “create a website” prompts.
Instead, you have to be specific. The more specific you are, the better your results will be.
Give it notes about what the output should be. Tell it which stack you want it in. Let it know your thinking. And be sure to give it the requirements for the app or website.
I’ll talk about this more in a minute, but a great way to start is to create and plan, then work with the AI to refine the plan. That way, by the time the plan is complete, you have a document for the AI to follow, with everything it needs to do.
You’re going to get out what you put in.
Speeds Up Writing Code
I have found that it definitely helps speed up the process of writing code.
For example, with Laravel, I like all my models to have controllers, policies, and requests —the whole nine yards. By hand, this can be a challenge. But with an AI, I can speed up the process, and while it’s running, I can think through the following steps.
Or, on a recent project at work, I needed five sets of models (and the works), listing pages, add pages, and edit pages. I created the first set and then told the AI to follow its example, with specific changes for each model.
The result was a process that took me less than an hour for all five, rather than manually creating everything.
When used correctly, AI can help speed up development. When used correctly, AI can help speed up development.
You Still Have to Analyze and Think Critically
Above all, AI does not replace critical thinking or code analysis.
I hope I’ve demonstrated that there will always need to be a human element to review code, design, and apps and websites as a whole to make sure they actually work.
You have to think critically about how your apps and websites are structured. You have to look at the actual code to make sure it makes sense. You still have to run tests, consider security, and ensure what you create is accessible.
AI does not replace your thinking.
Using AI for Planning
One really great use for AI in web development and coding is planning.
I’ve found that if I start by creating a plan with Claude, using my notes, defining what I want the end product to be, and adding a little Laravel Boost, I can get the result I want faster.
With the plan, I can see what the AI is thinking and catch potential issues right at the start, not after I’ve spent hours or even days working on the feature. Plus, the AI occasionally catches things that I miss.
The result is that I end up with something better than I would have if I had just given it a simple prompt.
Using AI for Actually Writing Code
And then of course there’s using it to write code.
I’ve found that Claude is the best for coding. It hits the right balance of speed and getting it right.
Gemini has gotten a lot better over the past year. Plus, their CLI tool makes it easier actually to use in PHP Storm.
Speaking of PHP Storm, JetBrains’ Junie started robust back in May and June, but has definitely fallen behind over the course of the year. It seems to take much longer to do anything than the others.
If you’re getting started in programming with AI, Claude is probably where you want to start.
Using AI for Tests, Fixes and Processes
Finally, AI has absolutely helped me with writing tests.
I know how to create tests and why they are essential, but I’ve always struggled to come up with the proper tests to ensure my code is as well covered as possible.
With AI, I can tell it to create a test suite for a particular feature, and it does it for me.
Also, it helps a lot with fixing code. A combination of Code Rabbit and Claude can help me ensure my code is clean and ready to merge before I merge it.
And I’ve been using it to help me get my processes in place, especially with the GitLab pipeline. Instead of hoping someone has the exact pipeline example that I want, I can tell it what I want for the pipeline, and it builds it for me.
If you want to start dabbling with AI in your development workflow, you might start by having it write tests, check for fixes, or audit and build processes to help you out.
Using AI for Content?
I promise you I am not using AI to write the content for me. I mean, I have a journalism degree. I went to school for writing.
I’m not going to let it take that from me.
Blog posts I write are all mine.
I might use AI to check my writing for mistakes or areas for improvement, just because I need a second pair of eyes.
I also use it to help write excerpts, titles, meta descriptions, and the occasional social media post for my blog posts, but it’s using my content, and I am generally not good at writing those.
I’ve also learned that using AI for deep research can be pretty good as long as you ask the AI to cite its sources with links and then check all of those links to make sure that a) they go to an actual page and b) that the stats are actually correct. You cannot passively use it and trust it.
As my first journalism professor said in class, “If your mother says she loves you, check it out.”
What are you using AI for?
So now I want to hear from you. What, if anything, are you using AI for?
Or are you refusing to use AI at all? That’s perfectly fine as well.
I’m curious to see where people land on this highly divisive topic.
