Why taste is the new bottleneck
When anyone can generate a hundred options in a minute, the scarce skill stops being production and becomes judgment.
For most of my career, the constraint was production. Making the thing — the mockup, the prototype, the working build — took time, and time was the wall everyone hit. AI moved that wall.
Now I can generate ten directions for a screen before my coffee is cold. The bottleneck didn't disappear; it relocated. It moved from making to choosing.
Generation is cheap, judgment is not
The uncomfortable part is that taste doesn't scale the way generation does. A model will happily hand me a hundred variations, but it can't tell me which one is right for this audience, this brand, this moment. That's still the work — and it's the part I can't offload.
What I've found is that working with AI exposes your taste rather than replacing it. When you can try anything, what you choose says everything.
What this changes about how I work
- I spend less time executing and more time editing.
- I write down why I rejected something, not just what I picked.
- I treat the model as a fast intern with no context — great at volume, useless at deciding what matters.
The skill ceiling didn't lower. It just moved to a place that's harder to fake.
I don't think this is bad news for designers. If anything, it's clarifying. The people who know what "good" means — and can defend it — are about to matter more, not less.