Every new tool promises to save us time. These ones do. But they also remove something that used to do invisible, important work — the friction that told us whether an idea was worth pursuing at all.
Before, if a blog post took three hours, you thought hard about whether to write it. Committing three hours of your life to something is a real decision. You made it deliberately, or you didn't make it. The friction enforced a kind of judgment by default.
Now the post takes eleven minutes. So you write it. Then the next one. Then the next. The friction is gone and with it the forcing function that made you ask whether the thing was worth making.
The filter moved
The filter didn't disappear. It just moved from execution to judgment. The question is no longer "can I do this?" but "should I?" And that is a harder question, with no model to answer it.
The "should I?" question requires taste. And taste, unlike execution, doesn't scale. You can generate a hundred variations of a thing in a single afternoon. You can only have one genuine opinion about which one is good. That opinion has to come from somewhere inside you. It has to be earned.
Anyone can now make anything. So taste is the whole game.
This is actually a more interesting problem than the execution problem was. Execution was learnable — most skills are. You practiced, you improved, you could measure the improvement. Taste is cultivated differently. It accretes slowly, through years of reading carefully, paying close attention to what works and what doesn't, noticing your own reactions and interrogating them. You can't shortcut it with better tooling.
What the shift costs
There is a version of this moment that goes badly. It goes badly when the unlimited ability to generate things is confused with taste, and the result is a flood of technically competent output that no one asked for and no one needed.
I see this regularly. The tell is a certain kind of eager shapelessness — content that could have been written by anyone, for anyone, about anything vaguely adjacent to the topic. It's fine. It means nothing. It accumulates. It takes up attention without paying anything back.
The tools aren't responsible for this. The missing judgment is.
What the shift offers
There is also a version that goes well. It goes well when the leverage is pointed at something specific — when someone with genuine knowledge of a narrow domain uses the tools to articulate that knowledge more fully, more clearly, more usefully than they could have without them.
That person doesn't produce more. They produce better, and more precisely. The tools are handling the mechanics of production, which frees the person to spend more time on the question of what's actually worth producing. That's the edge, and it's real.
The people I see thriving with these tools are the ones making less, not more. They use the leverage to compress the boring parts and spend the freed-up hours choosing what to point at. Not more output. Better judgment about which output is worth having.
How to cultivate taste in this environment
Three things have helped me:
First, consume deliberately. Read work that has survived — books, essays, journalism with a track record. The model trained on everything, including a lot of mediocre content. If you only consume AI-generated material, your reference points drift toward the average.
Second, protect the judgment step. Before asking the model to generate anything, write a sentence about what good would look like. Not the brief — that's a later step. This is a prior step: what are the standards by which you'll evaluate the output? Articulating that before you see the output is the only way to maintain it.
Third, say no loudly. The infinite supply of makeable things creates pressure to make. Resist it. The best practitioners I know are ruthless about what they don't produce. The no's are where the taste actually lives.
The tools are infinite. The attention you can give to the results isn't. Spend it on things you actually believe in.