Design with AI? Think different

With all the AI tools available to us, it seems that a designer has no need for the archaic, simple, not auto-completing, software of the past.

Why bother with something so trite as designing when AI can just do it all for you?
This week I've been having a conversation in group chat about this very thing, and how many of us in the industry notice the ai-ification of design across apps and websites.

On the one hand, I get it. Design is hard. It requires thinking, hammering away at an idea, chiseling until nothing else can be removed, and what remains is elegant and understandable.

Who wants to go through the trouble of that when you can just bid an LLM to do it for you?

This reminds me of when stock photos came into vogue decades ago. Every web designer began to resort to them, often not out of desire, but out of an easy solution to a problem—namely that website owners needed their domains to look interesting, and stock photos offered a veneer of sophistication to words on a page.
We've been through these patterns before. Fonts, design libraries, component libraries, images, copy. All have fallen prey to the mediocre average of whatever can get-the-job-done. I laughed to myself this month when I got a package in the mail from Amazon for a product whose labeling still had lorem ipsum on the instructions.

But, of course it's different now. We're not living in the same world as before—where lorem ipsum had its limits, only so many stock photos could exist, and the default bootstrap (or foundation) design elements had their limited usefulness.

Now, with a single prompt, you can will a website into existence. You can use Claude to build an entire mobile app for the iPhone, and have it fire up Xcode as well to run it in the simulator.

Anything can be made with words. And a lot of it is being created without any understanding of design. Now, I might sound like an old man shouting at the clouds. But hear me out. Design seeks to offer solutions, solves a problem, and makes life easier for an end user.

These—things—that get spit out from LLMs have no understanding of any of these needs. Instead they look at the corpus of slurped up data—ironically much of it from stock image laden sites themselves, and spit out approximations of design akin to the six-fingered men of the early days of LLMs.

And, here again, I get it. Design used to have this air of majesty to it—you'd hire some mystical creature to come down from the mountain, proclaim their words of wisdom, and ask for coin in place of worship (while accepting that as well).
That level of friction is just not acceptable for most people. Why go through all that trouble to work with some "artistic" type, when now you can just talk to ChatGPT and have it build something for you?

I've bemusingly begun counting the posters in drive-thru restaurants. In the last few months I've seen a shift from hand-drawn notes on the glass (as well as poorly formatted Word documents) to AI generated designs listing the food, catering services, menu options, or anything else that a manager wants to share. It's not great. But it's also not bad. Yes, it's horrible design. But it also conveys information in a way that's immediately useful.

Now, in the old days I would have been one of those designers that got paid to build out that poster—I did my fair share of those in the early days. I'm torn on that, torn on the fact that work previously given to designers is now thrown into the chat machines. But I also know that the quality of work I was capable of, and what actually got created—well it sometimes wasn't able to be much better than those ugly posters on drive-thrus.

So I've been weighing all this, and trying to figure out where we're headed. Right now, if you want to stay in the design field, or enter it in a post-GPT world, you need to look for ways to beat the machine at its game. You cannot (at least as of summer 2026) have Claude or ChatGPT or Gemini build a good poster, a great website, a well-thought app. It can do all of those. But they're janky, mediocre, and lack the conviction of a designer with a lived experience. Maybe that will be meaningless, but for now I have to believe it counts for something. So use these tools, understand how AI can help you, but then lean into your humanity to give us work that has opinions, leads, lacks deferential language, and is bold. I can't believe that AI would have given us the Think Different campaign.