When I sit down to use my computer, there are two things that I want to do. Either I want to use the computer for the sake of using the computer, or I want to use the computer to get something else done. I’ve spent the last couple of years using—and loving—macOS. In the past couple of years, I’ve never had an OS get out of the way and just let me work quite as well as macOS does. Not Linux, and certainly not Windows. But in the bliss that comes with being able to ignore an OS, I feel as though I’ve begun to lose touch with what got me interested in computers in the first place—an endless love of tinkering.

Compared to the endless options that Linux presents, macOS is comparatively lackluster. Sure, I can change the wallpaper or maybe configure a nice theme for my terminal of choice, Ghostty, or change where the dock resides. But if I wanted to truly change how the UI or UX of macOS works, then I’m SOL without some more serious tweaking. Before moving over to macOS, my distro of choice, EndeavourOS, and my DE of choice, KDE, offered nearly endless ways to change both what I saw and how I fundamentally interacted with the OS. And that’s something that I miss, even if I don’t want to change anything on macOS and spent most of my time with Linux just making it look as close as possible to macOS. I think what I miss is knowing that, at a moment’s notice, I could let the ADHD take the wheel and just spend a few hours reconfiguring how my desktop looked and worked.

So why don’t I move back to Linux? The Asahi project, while they may have had a couple of setbacks in the past year, looks amazing. And I do have an M2 MacBook Pro, so I could make the switch. The biggest thing keeping me on macOS really is just that I have things that I have to do on my computer. I’m a college student, I have a job, and I love spending time with friends away from the keyboard. To put it simply, I’m not spending as much time with my computer as I used to, and when I am, I’m either wanting to relax and play some games or I’m using the computer to get something else done. Even in my CS or cybersecurity classes, I’m either coding or using a VM—to say nothing of my poli-sci classes. I’m still using the computer to get something else done.

For full disclosure, I wrote the prior few paragraphs a couple of months ago, and am just coming back now, in July, to add to this post and clean it up. And my stance has kinda just become more concrete. I fucking love macOS, I really do. It does just work. But, at the same time, it rarely breaks or has me doing anything interesting, and so yeah—it has made computing less fun for me in the name of “It just works.” macOS Golden Gate continues that legacy of just getting out of my way. Many of my complaints with how Tahoe worked have been addressed, and I find myself enjoying using a system in such a way that I have scarcely felt before!

And then there’s AI

I have to admit, I’m not an anti-AI absolutist. I’ve used AI plenty over the past few years. Shit, I pay for my AI—not in monthly subscriptions, but rather in API credits, as I use a self-hosted LibreChat instance (a more private solution). In fact, I was really early on the AI curve. I remember back in 2021, when I was still in high school, playing around with very primitive AI after discovering it via the YouTuber DougDoug when he was using it to rewrite his LinkedIn profile using the now-defunct InferKit, and even having a bit of fun using it to help write a paper in my Earth Science class. Of course, I had no idea what it was all to become—I purely saw it as this silly little thing that you could get to say some pretty funny stuff. A couple of years later, I saw it as a potentially really helpful tool after watching a Tom Scott video about the burgeoning technology. It wasn’t until the past couple of years that the full breadth, and potential danger, of AI became fully evident to me.

But, for this post, I want to set aside everything else about AI (I plan on making follow-up posts addressing AI specifically and my thoughts on it) and just focus on how it makes me feel about doing computing tasks. And, in short, it kinda takes the fun out of it. Because it does everything itself. There’s no thinking, or at least very little, involved on my part. I think the perfect encapsulation of this sentiment is a recent project that I’ve been working on for the past few weeks with a friend of mine. We’re working on getting an RP2040 to output to an old Apple CRT. And it does. We got it to work. But… the whole thing is vibe-coded. Shit, we even used AI to tell us what to solder together. That’s the only real thing that we had to do, and the only real brainpower that we had to put into the project. Other than that, the whole thing—all 19k lines of code (mostly C)—are AI-generated, mostly by Gemini with a little bit by Claude.

I feel like I should be happy about the project. It looks cool as fuck, and it’s objectively a fun project. And yet… I can’t help but feel almost guilty about it. Or at least not quite as happy and accomplished as I might otherwise feel if we had chosen to forgo such heavy AI usage. I think what’s really driving this feeling is that, if we’re just gonna keep it a buck, I haven’t really learned much from this project. I didn’t really have to experiment or tinker a lot. All I did was tell a robot what I wanted it to look like, and sit back on the couch. With some of my past tech adventures, I’d have to spend hours delving into docs, trying different things, breaking a lot of things, and just… tinkering. Now I open up Antigravity or VS Code, punch in a paragraph about where I want the project to go, and sit back and watch as the AI does everything that I used to have to do myself. And, to be fair, it is doing a better job than I would be doing. It’s also taking out a lot of headaches and frustration. But it’s doing so by taking out, well, everything! There’s a certain magic, that I’m sure many will relate to, where you’ve been hyper-focusing on something for hours—forgetting to eat, sleep, or drink—and nothing’s working. Until you happen upon a Reddit post from 15 years ago describing your exact problem and, lo and behold, it works! And that provides enough dopamine and adrenaline to get to the next “Eureka moment.”

And I don’t know what the solution is. AI, objectively, makes everything about the development process a lot faster. And it objectively does a much better job than I could hope to do. But it does so at the expense of everything that made me want to really get into computing when I was growing up. I truly don’t know how the future generation, which has never experienced a world without AI, is expected to find the same level of love for computing and exploration that past generations had. Maybe schooling can help to foster it? But once you get out into the “real world” and get a job, that notion is gone. I have numerous friends working or interning in the industry, and from what it sounds like, most of their day is spent prompting whatever Anthropic’s latest model is rather than spending time writing lines of code or exploring codebases themselves.

Honestly, I don’t know that there is a good answer. For myself, I’ll probably continue to use some AI—it’s too useful of a tool to simply ignore—but then try to understand the project more intimately than required. But it’s hard sometimes when I just want it to “work”—AI is a very effective shortcut to that end. But I also want to have fun, and sometimes that means breaking shit and getting my hands dirty.

(AI generated writing and art can fuck all the way off btw)