I use AI tools for development (in the sense of coding things), but not for much of anything else. I like to keep up with what these systems are capable of, the leverage they provide users, but I wouldn’t apply them to my writing and am not personally interested in using them to create images, videos, and so on.

The distinction between what’s AI-augmental and what’s not, for me at least, is that I want to pick up just enough programming knowledge to help me accomplish things—to build little tools that I’d like to have exist as quickly as possible so I can use them—while writing and visual art/design are fulfilling and useful to me as practices; I’d like to continue performing them for their own sake, and I’d like to keep getting better at them.

At their best, these AI tools are really, really good at helping even halfway competent people churn out halfway decent end-products. If you know the basics of some field, you can probably produce stuff that’s a fair bit beyond your current capacity using said tools.

If you’re really good at something, you can probably generate a lot more volume than you would otherwise be capable of by adding AI into your process, and even if the quality won’t always be comparable to what you could do solo, on your best day, at times “quantity has a quality all its own.”

That distinction between doing to make and doing to do is important, I think.

I write to produce writings, but I also write to think through problems, improve my communication skills, and learn how to tell better stories. I draw, paint, and design to create drawings, paintings, and designs, but also to relax, to exercise my creative muscles, and to mess around with ideas that would otherwise distractingly jumble around in my head.

When I’m building a website or app, though, I usually just want the website or app to exist. And though I tend to pay close attention to what the AI is doing when I use such a system to help me build something, I don’t learn nearly as fast and thoroughly as I would, had I worked through the problem myself.

I’m not a fan of the companies behind these AI tools and a lot of what they’re doing. Their incentives are near-universally bad, they’re run by sociopaths, and they’re getting away with a lot of illegal stuff because they’re too big to punish and because they have more money to throw at legal fees than basically anyone else (including, at times, the governments who might otherwise hold them accountable).

That said, I’m also not vehemently anti-AI. Many of these systems are just tools that, if used appropriately, can be just as empowering (though in different spaces and via different means) as a hammer or calculator or Photoshop.

I think the real question as to how and when and whether we should use AI comes down to motivation. Am I trying to learn? Or am I just trying to churn something out?

For the learning stuff, it’ll usually make more sense to avoid having a piece of software do the labor for us. Unburdening ourselves of the frictions that would, over time, polish our capabilities into more refined shapes is a reliable way to deny ourselves the very upgrades we’re hoping to attain (or if we’ve already attained them, to over time lose those capabilities and that knowledge—to de-skill).

For the stuff we just want to wish into existence, though—essentially rubbing a lamp and asking a software genie to make us an app with a specific set of features—these tools work decently well. Though again, they tend to work even better for folks who have done some (or a lot) of learning, first.

If you enjoyed this essay, consider supporting my work by becoming a paid subscriber, buying me a coffee, or grabbing one of my books.