More options, fewer outcomes
A prompt is supposed to open up options. Used by enough people, on the same handful of tools, at the same time, it closes them.
What the model is actually optimizing for
A generative tool doesn't produce the most original answer to a prompt, it produces the most probable one: the output its training data suggests is most likely to be considered good. That's not a flaw, it's the mechanism. But it means the fastest way to get a usable result from these tools is also the way most likely to land near the center of whatever "good" already looked like across everything the model was trained on. Ask a thousand designers to generate "a clean modern dashboard" and the model isn't drawing on each of their individual taste, it's drawing on the same aggregate of dashboards it's already seen enough times to consider representative.
Default drift
Every design tool has always had defaults, and designers have always had to actively resist them to make something distinctive: the default font, the default spacing, the default template a client swears looks "clean" because it's the seventh version of the same clean they've seen this year. What's different now is the scale and the speed. A default used to spread through a plugin or a popular template, over months, to people who'd chosen that tool deliberately. A generative default spreads through a prompt, instantly, to anyone typing a few words, most of whom never see the hundred other outputs converging on the same look at the same time.
The aesthetic center of gravity that forms when enough people take the fast, plausible answer instead of pushing past it is default drift, and it moves faster than any design trend ever has, because it isn't spreading between people anymore, it's spreading between a person and a model that already contains everyone else's fast answer.
Where judgment actually has to show up
None of this is an argument against using these tools. The boring, mechanical parts of a workflow are still exactly where they earn their keep. It's an argument about where a designer's actual job moved to. The brand and product work I've done for creative studios lives or dies on looking like nothing else in the category. That's exactly the kind of work I'm most careful about running through a generator, not because the tool is bad, but because the entire point of that work is to not land near the average of what everyone else already tried. The job isn't generating an option anymore. It's recognizing when the fast option is the average one, and knowing that's precisely the moment to put the prompt down.
A prompt doesn't close your options because the tool is limited. It closes them because it's very good at giving you the same option everyone else is already getting.
