Knowing-How-to-See

From the hand to the eye: the decisive skill in the age of artificial intelligence

Our entire civilization of work is built on the praise of the hand. Knowing how to write, to code, to sell, to manage, to produce: we have learned to measure a person’s worth by what they are capable of doing. Execution was the proof. The deed attested to the skill.

Yet an older and more silent faculty is about to become decisive. It resides not in the hand but in the eye. It consists not in producing but in perceiving. I will call it knowing-how-to-see — and I believe it is this, from now on, that will separate those who matter from those who merely bustle.

Two faculties, not one

Knowing-how-to-do is a relation to action. It is learning to change gears, to parallel-park, to string gestures together, to follow a procedure. The Greeks called it technē: the art of bringing a thing into being, of drawing it out of nothingness into existence. Knowing-how-to-do answers the question how to produce.

Knowing-how-to-see is of a different nature. It fabricates nothing; it distinguishes. It recognizes what is true and what is merely conventional, what is dense and what is hollow, what has tension and what has none. The Greeks had a word for this too: krisis, the act of cutting, of separating, of discerning. It is from there that our words critique and crisis come, for to judge is always to make a division. Knowing-how-to-see answers another, harder question: what deserves to exist?

Knowing-how-to-see is often confused with taste, and taste with a subjective whim. That is a mistake. The seeing I speak of is not a mood, it is a competence. The photographer does not merely like or dislike an image: he sees that a light is flat, that a frame is poor, that a composition does not hold. He can objectify why. His taste is not a vague opinion, it is an informed taste: built by attention, by comparison, by repeated demand. Kant already said of aesthetic judgment that it lays claim to the universal without passing through the concept. Knowing-how-to-see is exactly that: an educated perception, able to justify what it discerns.

The fundamental asymmetry

Here is the decisive point, the one that changes everything.

One can have a knowing-how-to-see superior to one’s knowing-how-to-do. The fan analyzes the bad call, the absurd positioning, the pass made too late, without being able to reproduce a single one of those gestures himself. The dinner guest recognizes a dish that is over-salted, poorly balanced, without relief, without being able to make it again. Seeing does not require knowing how to do.

But the reverse is almost impossible. One never produces, durably, above what one is able to perceive. Perception is the ceiling of production. Whoever does not see what makes a sentence excellent, an architecture sound, a decision fine, may work for twenty years: his best work will remain average; and, worse still, he will not know it. For the eye is also what measures the gap between what one has done and what one ought to have done. Where the gaze is weak, no gap is perceived, hence no correction is possible, hence no progress. Incompetence is blind to itself.

It is a hard idea, but it is the key to everything that follows: the quality of what you produce is bounded by the quality of what you know how to see.

What the machine augments, and what it leaves intact

Artificial intelligence is a prosthesis of knowing-how-to-do. It is there, and only there, that it is prodigious. It raises everyone’s floor of execution by a few points. The one who wrote at a five produces at a seven. The one who could not structure a presentation comes out with a decent one. The one who could not code gets a prototype. The machine gives speed, form, a first version. It makes the doing abundant.

But it does not give the gaze.

And this is where the trap is set. The machine does not produce obvious ugliness, it produces plausibility. A smooth text, an assured syntax, a clean layout, a reasoning that has the look of reasoning. To the untrained eye, this plausibility is indistinguishable from quality. We believe it is good because it is well presented. We take ease for depth, and fluency for thought.

The same document, placed under a trained eye, at once reveals its flaws: this passage is empty, this idea is conventional, this image is pretty but without intention, this strategy does not hold beyond the first sentence. And above all, that gaze knows how to ask for better. It steers the machine, corrects, tightens, demands. It turns a tool into a lever. Where the weak eye settles for the first answer, the demanding eye extracts the tenth.

The inversion of value

A simple law governs all value: what becomes abundant becomes cheap. And AI is precisely in the process of making knowing-how-to-do abundant. Producing, formatting, synthesizing, generating: all of this will become accessible, assisted, almost free. Scarcity is therefore leaving the hand.

And when a resource ceases to be scarce, value migrates toward what still is. Scarcity does not disappear; it moves. Tomorrow it will reside in judgment. Not in the capacity to produce, but in the capacity to recognize, to choose, to prioritize, to formulate a demand, to see the difference between a correct answer and an excellent one. The differentiator will no longer be the deed. It will be the eye.

AI does not, then, abolish human demand. It shifts it from the hand to the gaze, from doing to seeing.

The real risk

We keep saying that the danger of these machines is that they produce mediocrity in bulk. That is true, but it is not the real danger. The real danger is more insidious: that we lose the faculty to recognize mediocrity.

The gaze is a muscle. It is maintained by friction, by effort, by comparison, by repeated judgment. And we are entering a world where everything will look professional, where everything will be smoothed, formatted, instantly available. In such a world, the exercise of discernment becomes at once more necessary and more rare. It is a vicious circle: the less the eye is called upon, the less it sees; and the less it sees, the more plausible mediocrity prospers unseen. The faculty atrophies at the precise moment it becomes vital.

That would be the true defeat. Not machines that produce too much, but human beings who no longer know how to judge, who take abundance for richness, and the passable for the excellent.

What remains

Knowing how to see is thinking with precision. It is having a trained taste, capable of ranking quality in a universe where everything looks alike and everything presents itself as good. It is refusing the first answer because one perceives the one that could have been better.

When knowing-how-to-do becomes accessible to all, knowing-how-to-see becomes the only true differentiator. The future will belong less to those who know how to produce than to those who know how to recognize what deserves to exist.

And perhaps that is the last irreducibly human gesture: when the machine can do everything, it will remain for someone to decide whether it was worth doing.

#chapter1