Lately there’s been endless discourse about AI in Michigan. Mostly about data centers, their development, their energy use, and the political corruption involved in their contracts. Lost in the weeds is the premise that AI could prove a boon to Michigan, but not for the reasons anyone thinks.
The biggest reason? AI coding tools are making software development accessible to anyone, without needing to learn complex code.
The early adopters call it vibe coding. Using tools like Claude Code or Cursor to write code for you, without needing to code things yourself. You simply dream up an idea for an app, and Claude makes it for you.
What it might lack in precision, it makes up for with accessibility. If you can type in English, you can interface with tools like Claude Code. Basic computer knowledge is definitely helpful, and basic programming knowledge certainly helps make outcomes more specific.
The real promise of AI tools is making computers more relational. The interface itself is revolutionary. You don’t need to use a command line, or even a graphic user interface, like modern computers use. Just plain English.
I’ve always been fascinated by computers and have what I’d consider an expert-level amateur knowledge. I never got seriously into coding, beyond basic web development. It seemed cumbersome and difficult, abstract and obtuse.
Most people feel that way, I’d imagine, which is why software developers still command such high salaries. It’s an incredibly difficult skill, one which most people either don’t want to learn or aren’t intellectually capable of learning.
Vibe coding changes that. Take me, for example.
Often in my line of work, I have to download music or a web video clip to cut into a short-form video. There aren’t many publicly available tools to do this, and sites like YouTube don’t allow you to download content directly.
You have to really search around and find a downloader tool. The best one I’d found was a command line interface called yt-dlp. It forced me to boot up the command line terminal on my Mac every time and remember a slew of text commands to format the downloads how I wanted.
I’ll admit, firing up the command line made me feel like a serious programmer, but it became so tiresome. I always forgot the specific commands, or mistyped them, and it took me way longer to download things than necessary.
My first project with Claude Code—make an application for yt-dlp, with a proper graphical user interface. A visual app for my Mac, a program with buttons and toggles that was easier to use, with all the commands baked in, so I could download videos without wasting time.
Claude spooled it up for me in just a few minutes. It even made the logo, using my idea. After some testing and tweaking, all done by just typing back and forth to Claude, my application was ready. The whole thing took less than an hour.
That was a simple application, but since then, I’ve developed a slew of others. My favorite, a custom photo editor for an art project that I wanted to build years ago but was too complicated to code myself.
In every case, I opened Claude Code with an idea, and it just… made it. The programs weren’t perfect from the start, but the software was there and just needed to be tweaked and debugged, all done just by typing with Claude, and completed in an afternoon.
More sophisticated programmers, the early adopters, are using it for even more complex projects. Spooling up dozens of Claude Code instances at a time, having each AI “agent” independently work on tasks, and managing them as they go about it.
The irony is, the first thing people think when they realize the potential of AI coding is that they could use it to make software and sell it for a bunch of money. That’s been the model for a generation, where software programming was a highly sought-after, gatekept skill.
That’s increasingly no longer the case, however. In a world of easily generated AI software, basic programs are no longer valuable enough to be sold. This forces us to reimagine software entirely.
The real potential of vibe coding is making bespoke apps for your own use, not apps to sell to others.
You can find that annoying thing you have to do at work every day, that annoying pain point for your business, and just make a program that makes it go away. Without needing to pay tens of thousands of dollars to a consulting company to make it for you.
The paradigm for software development will increasingly shift from a model of building programs to sell to others, to building programs to use yourself.
Think about it more like carpentry. You still need advanced, highly skilled carpenters for complex jobs, but just about anyone can hammer a few nails when needed. That’s where software development is headed.
For Michiganders already used to doing things independently with a characteristically hardworking, entrepreneurial spirit, it will prove revolutionary.
Of course, there are limitations, even as the tools develop. Michigan farmers won’t be vibe coding complicated financial software anytime soon. But, they could easily make a program that monitors networked milking machines for their cows, or a bot that tracks the price of corn futures.
Not years from now, not tomorrow, but right now. That capability exists right now, but the early adopters, permanently online obsessive coders, are the only ones seeing it.
They have a big first mover advantage, but it won’t last forever. Claude Code only costs 20 bucks a month, after all. There are other similar AI coding apps as well. The space is brimming with competition.
So yes, when you hear about datacenter development in your backyard, you’re right to be suspicious. It should be done fairly, with local job hiring, and energy production built up without costs passed to general consumers. Politicians (and their spouses) shouldn’t be involved in corrupt deals to build them, either.
But in the end, we’ve made a civilizational pact with technology, unless you’d like to go be Amish. Data centers will be built somewhere, somehow. It’s our responsibility as citizens to make sure that happens fairly, in a way we benefit from, but it’s something to be encouraged, not feared.
The opportunity for individuals employing advanced AI tools, like these coding bots, is simply too great to pass up. It could prove a true generational leap, completely redefining the way we interface with our technology, making it more accessible than before.
This is the real potential of AI, far beyond the chat bots and video slop. That’s the most visible output presently, and yeah, it sucks. But for makers, for people who build things, we’re at the tipping point where AI tools have actually changed the game.
We need more data centers to build it. Maybe it’s time to stop fearing them.
Bobby Mars is art director of Michigan Enjoyer. Follow him on X @bobby_on_mars.
