Why I Don't Think AI is the Devil
I’m a technology enthusiast, not a business analyst, so I’m not the right person to predict what all the money pouring into AI is going to do to the economy. I consider myself an environmentalist, but I’m also skeptical that every scary headline about AI data centers and energy consumption is entirely free of Luddite panic. I don’t love some of the content AI produces, and I understand why people call it a plagiarism machine for the way it borrows from the work of others without credit.
My own relationship with AI started pretty simply. I began paying for Google Gemini in 2024, and when I finally de-Googled last year, I moved over to ChatGPT. This has been my first year of retirement, and I spend most days at home writing and teaching myself the technologies I never had time to dig into while I was still working. I built a self-hosted server. Then I built a better one. AI ended up helping me constantly with configuration files, scripting, and general Linux best practices. For months, that was almost all I used it for.
As I’ve gotten more comfortable managing my setup, I’ve needed fewer basic operating-system answers. That’s pushed me to explore other ways to use large language models—things that play more to their strengths. Here are a few that surprised me.
Nutrition
I’m diabetic and my wife has celiac disease. We wanted to follow the Mediterranean Diet for health reasons, but finding something that fit both of us felt overwhelming. I asked ChatGPT to create a meal plan that accounted for our restrictions while still sticking to the spirit of that way of eating. It produced a menu that actually made sense, and then generated a shopping list I could take straight to the grocery store. That’s not earth-shattering technology, but it was genuinely useful.
Twelve Years of Crud
I’ve used Migration Assistant on every Mac I’ve owned since 2014. That means my keychain still had passwords for Wi-Fi networks that haven’t existed in years. My shell configuration file had become a digital junk drawer filled with leftovers from every command-line utility I ever experimented with. My PATH statement was longer than my last will and testament.
With ChatGPT’s help, I methodically cleaned out a lot of that accumulated cruft. It helped me identify obsolete entries, simplify scripts, and generally untangle years of well-intentioned mess. A few persistent little bugs disappeared along the way. That alone made the subscription worth it.
Creating a Reference Library
After a few months of leaning on AI to solve real problems, I realized something embarrassing: I hadn’t taken good notes. I’d solved issues, found clever scripts, and learned useful tricks—but I hadn’t organized any of it in a way I could easily find again.
I downloaded transcripts of all my conversations with ChatGPT and fed them back into it to build an index. From there I converted everything to Markdown and dropped it into my Obsidian vault. The AI helped generate tags and backlinks so the notes connected to each other. Now, when I need some obscure command or script, I can actually locate it instead of trying to remember which rabbit hole I went down six months ago.
Task Management Redesign
I’ve borrowed ideas from David Allen’s Getting Things Done system for years because the basics are solid. When I think of something I need to do, I write it down immediately. I do periodic reviews. I try to organize tasks by context. But I’m not a GTD purist, and I don’t want to live a life ruled by checklists.
I use Things 3 as my task manager, and over time it had quietly turned into a disaster. I had abandoned projects that kept accumulating tasks, overlapping tags, and—somehow—duplicate folder structures that I kept maintaining out of habit even though they made no sense.
Eventually I dumped the whole mess into ChatGPT. I listed my tags, areas, projects, and all the quirks of my broken system. It came back with a simplified structure: six clear areas, a couple dozen sensible tags, and practical guidelines for deciding when something should be a real project instead of just a task with a checklist. It even walked me through setting up an integration with Apple Reminders and Siri so I could add tasks by voice—something I’d always meant to do but never made time for.
Conclusion
ChatGPT is far from perfect. When I ask about software, it often relies on outdated information and gives instructions that don’t quite match the current version. Sometimes it veers off into ridiculous territory, like suggesting I could use a phone app to control a remote Mac in ways that clearly aren’t possible. It can get weirdly self-referential and sound like it’s pretending to have lived human experiences. And even as an editor, it occasionally suggests changes that miss the mark.
But despite all of that, it has saved me an enormous amount of time and solved problems that used to take hours of research and trial and error. I see it as the next logical step in the evolution of everyday computing, not the end of civilization. It’s just another tool—one I use with a healthy dose of skepticism, the same way I’ve always treated tech-forum advice from someone named bIgwORm007 confidently explaining how to overclock a CPU with “facts” that may or may not be entirely imaginary.
Enjoyed it? Please upvote 👇