Slow Is the Work

I caught myself rushing to iron out edits on a screen that had no right existing. I hadn’t thought it through.¶

This is the AI version of a habit I’ve watched a lot of designers fall into. The instinct used to be to sketch through the problem until the answer surfaced. Now the instinct is to skip the sketching and start polishing. Generate something. Iterate. Ship. The motion feels like progress. The output looks competent. The work doesn’t get better, but the calendar fills up with shipped things, which is the part everyone can see.¶

AI is fast thinking. It’s a pattern-completion engine. It hands you plausible output in seconds, and the output is usually fine. That speed is the gift. The same speed is the trap.¶

The trap is that the model’s speed pulls your speed up too. You stop sitting with the decision because something good enough is on the screen, and fast is better than perfect. The hour you would have spent wrestling with the problem becomes the minute you spent picking between five outputs the model gave you. The output looks fine. It IS fine. But fine doesn’t compound into great.¶

This is where AI slop comes from. Not the model. The tempo.¶

I’ve shipped some of the best work of my career with AI in the loop. What separates that work from slop isn’t the tool. It’s whether I let the model’s speed become my thinking’s speed.¶

Design isn’t pattern completion. It’s pattern recognition. Looking at a working thing and seeing what’s wrong before you can name why. Changing your mind. Changing it back. Sitting with the version you don’t love long enough to figure out whether the problem is the version or the assumption underneath it. That process is the work. AI didn’t take it away. It made skipping it feel productive.¶

The slow part is where the decision actually gets made. Most of design at any altitude (craft, system, product, strategy) is the same loop: produce a thing, sit with it, notice what doesn’t hold, fix the root, not the version. AI compresses every part of that loop except the sitting. And the sitting is the part that decides whether the thing actually works or just looks like it could.¶

AI takes me to a working version of the thing. I take it into Figma to find out what’s actually wrong with it. The canvas is where the slow part lives now. Alone, focused, with the file pulled up next to the running product, editing decisions I made too fast the first time. The cut happens after the generation, not during. The cut is where the judgment lives.¶

Sweat equity isn’t nostalgia. It’s the part of the work that doesn’t show up in the output but shows up in the result. The hour where you wrestle with a decision, settle, and walk away knowing you actually decided. That feeling isn’t ornamental. It’s what’s missing from work that looks fine and isn’t.¶

The story most people are telling about AI in design is about speed. Faster mocks, faster iteration, faster decisions. That story names what’s possible. It misses what’s required. Speed is the easy half. The hard half hasn’t changed. You still have to sit with the thing until you actually know it.¶

Speed is the model’s job. Slow is yours.¶

The slow part is the work.¶