There was a moment, watching the tools get genuinely good, when I felt the ground shift. Things that used to take me a day now took a sentence and a few seconds. Building — the thing I’d spent years getting skilled at — had suddenly gotten cheap. And my first, honest reaction was to wonder what that made me worth.
The answer took a while to arrive, and it turned out to be reassuring. When building was hard, the ability to build was the scarce, valuable thing. Now that a machine can produce almost anything you can describe, the bottleneck has moved somewhere the machine can’t follow: to knowing what’s worth building in the first place.
That’s taste. The sense of which of the ten generated options should actually exist. What to leave out. What “good” even means for these particular users, in this particular context. The model can hand me endless possibility; it can’t tell me which possibility is right. That judgment is mine, and it’s suddenly the expensive part.
And taste isn’t something I could download even if I wanted to. It’s the slow residue of years of seeing what works and what quietly fails, of caring about the difference, of being wrong enough times to develop an instinct. As production drops toward free, that hard-won judgment is going the other way — becoming the rare thing.
So I stopped mourning the skill that got cheap and started deliberately sharpening the one that got precious. When everything is possible, the person who knows which thing is worth making is the one who matters. Taste is the part of the job that’s still, unmistakably, mine.
– Serguey Asael Shinder
Leave a comment