Everyone's using AI. A lot of it is embarrassing. I use it anyway, and here's what I actually think about it.
There's a version of this article that opens with a hot take about AI replacing designers, or a reassuring counterpoint about how creativity can't be automated. I've read about forty of those in the last year. This isn't that. This is just what I've actually watched happen, from inside it.
Something shifted when companies discovered AI, and not entirely in the right direction. The obsession moved fast - automate everything, measure everything, show velocity at all costs. Weekly summaries of lines of code written started appearing as proof of productivity, as if volume was the point. Every process became a candidate for automation, regardless of whether automating it actually made sense. People started defaulting to the most expensive models for tasks that didn't need them.
And then there's the content. AI-generated decks, AI-generated briefs, AI-generated copy that covers every expected point and says nothing that anyone needed to hear. There's a particular texture to it - technically complete, completely forgettable - and people have started to feel it even when they can't name what's bothering them. The volume of output went up and the signal-to-noise ratio collapsed. Everyone is producing more and communicating less.
This is what AI-as-performance looks like. Impressive on paper, questionable in reality. And nobody wanting to be the one who says it out loud.
Because I use AI every day, and I'm not going to pretend otherwise or frame it as some reluctant compromise. The honest version is this: I use it because I want to survive in an industry that's moving fast, and because it genuinely lets me build things I couldn't have built otherwise. Those are two different reasons and both of them are true at the same time.
The survival part is straightforward enough. The market has decided that people who can move across disciplines are worth more than people who can't, and keeping up with that is just how things are right now. But the other part matters more to me personally. Thanks to how these tools work now, I can take an idea from concept to a working product without an engineering team, without convincing anyone it's worth prioritising, without waiting for a roadmap slot that never comes. I built Trading Charts. It's a Figma plugin, nothing groundbreaking, but before these tools existed, it would have required either an engineer or a level of patience I don't have.
There's also something unexpectedly satisfying about running an entire product by yourself, knowing you can take something from start to finish. Even if some of the code is fragile in ways a real engineer would wince at. It works, people pay for it, and I learned more building it than I did in years of designing for other people's products.
People can tell. Not always consciously - they don't open a website and think "a language model wrote this". But they feel the absence of something. A perspective, a real decision, a person who actually gave a damn about what they were making. AI-generated content has a texture that audiences have developed a sensitivity to, even without being able to explain it. It covers the topic, hits the expected points, and doesn't say anything that surprises anyone or makes them feel anything.
The work that's clearly made by a person still gets shared, still gets remembered, still creates the kind of connection that makes someone come back. Genuine craft is actually more valuable now than it was before all of this started, because there's so much more noise to cut through.
It can't be curious, and it can't care. It can execute on curiosity very well, which is useful, but the curiosity itself has to come from somewhere real. Every good project I've worked on started with someone staying bothered by a problem longer than was comfortable, trying something that probably wouldn't work, finding out it didn't, and going again. You can't prompt your way into that.
Garbage in, garbage out is the most accurate thing anyone has said about these tools. If you don't have taste, if you don't have a real point of view, if you're not bringing genuine curiosity to what you're working on - the model just gives you more of nothing, faster. The output is a direct reflection of the quality of thinking behind the prompt, which means the people who are going to do well here are the ones who were already good at the thinking part, not the execution part that's being automated away.
I still fucking love doing this manually, especially when it comes to designing. The craft is still there, the curiosity is still there, and the need to get something right, to go back and fix the thing that's bothering me even when it's probably fine. That's entirely mine and I don't want to give it up. AI extended what I can do. It didn't change why I do it.