Taste is the thing AI cannot borrow
Models are very good at the average of everything. That is precisely the problem.
Category,
AI
Tomas Lindqvist
,
A model can produce a competent version of nearly anything. What it cannot do is care that the competent version is the wrong one.
Averages are not positions
Generative output converges on the statistical middle of its training data. For most design problems the middle is exactly where you do not want to be, because your competitors are already there and the whole point of the work is to be legible as distinct.
This is not a limitation that better models fix. Fluency and judgement are different capacities. More fluency produces more plausible options, faster, which raises rather than lowers the value of knowing which one to choose.
The skill that appreciates
If producing options becomes cheap, then selecting, editing, and defending a point of view becomes the scarce part. That has always been the senior skill. It is now the whole job.
