We’ve seen code generators becoming more integrated into the tools we use every day. Some call them "the future of coding," others are more skeptical and stick to using Vim, I'm in the middle and I believe there's a strong similarity between AI generated code and WordPress.

WordPress powers about 43% of all websites on the internet, including some of those shiny Fortune 500 companies' sites. It’s an absolute workhorse for blogs, small businesses, and even some more complex sites. But we all know it's not ideal for everything, especially when you're aiming for real scale or complexity.

WordPress: Great Tool, Wrong Job

Doing a quick rundown of things WordPress isn’t exactly designed for:

• Large-scale platforms (Think millions of users simultaneously hammering your server)
• Complex web applications (Like heavy SaaS with intricate workflows)
• Real-time collaborative tools (Imagine building Google Docs with WordPress... yikes)
• Financial platforms (I’m talking about high-security requirements, not a simple eCommerce site)
• Mobile-first apps (Yes, you can build apps with WordPress, but should you? Let’s be honest)
• Data-heavy analytics platforms (Good luck crunching massive datasets on WordPress)

The list can go on, WordPress is amazing for what it is – just not for everything. It’s great for certain applications but definitely not the go-to when you’re looking at high security, real-time functionality, or complex operations.

AI Code Generation: Familiar Territory

In the same vein, AI code generation or IDE assistants (take your pick: Copilot, Tabnine, Cursor, or any fork of VS) are fantastic tools that can really streamline your workflow, but they’re no silver bullet. Just like WordPress, they’re wonderful when used for the right task, but don’t expect them to build your next complex, highly-secure, real-time platform without some serious oversight.

The code they generate isn't always the best. We’ve all seen those StackOverflow snippets pasted verbatim that work (sometimes) but make you wince in horror when you peek under the hood. The logic these tools implement can often be faulty and it make take you more time to code-review the generated code than to just write it yourself.

Where AI Shines: Scripting and Simplicity

AI truly shines in automating repetitive scripting tasks or creating "black-box" logic. The kind where you feed in some input, expect an output, and don’t really care about what happens in the middle? As long as you can define the task clearly, AI can often hit the nail on the head.

If you’re looking to automate a tedious process, spin up a basic REST API, or even mock up an MVP (with very simple functionality), AI code generation might just save you hours of work as I've shown in my article Generative AI: Dynamic Mock APIs using Langchain and Anthropic..

I've also used it to generate tests, I've found that it does a pretty good job at a task that is usually quite boring - UNLESS you're doing TDD, which in that case I would not recommend it.

Pro tip: the more you can reduce your use cases and make your MVP as basic as possible, the higher the chances you can get away with using AI-generated code. Less is more here.

The Silver Lining

Just like WordPress, AI-assisted code generation is a tool in your toolkit. It's powerful, but only when used correctly. Think of it like trying to use a hammer to do the job of a screwdriver – sure, it might work in some scenarios, but you're probably going to break something if you push it too far. 🛠️

At the end of the day, tools like AI code generators can take a lot of the grunt work out of your hands and speed up development for specific tasks. But when you start needing more robust, nuanced, and scalable solutions, that's when the human touch (aka: you) is irreplaceable. AI won’t be putting us out of a job just yet.