Like This, Not Logic

Disney does it. A five-year-old does it. Spinoza did it. AI cannot.

A five-year-old shows you a flower. She doesn’t classify it. She points. She says, “look, like this.” You look. You see what she sees.

This is the gesture I want to write about. It is older than language. It is the actual primitive of teaching, of showing, of one mind landing on the same thing as another. And it is what we keep failing to notice in our discussions of intelligence.

Disney’s whole storytelling apparatus runs on it. No Disney film argues a moral. Disney shows you the moral, in pictures, with characters you can imitate — the prince who learns to love what looks broken, the daughter who walks back into the burning castle for her father. You don’t conclude anything. You see something. The seeing is the teaching.

Spinoza, three hundred and sixty years ago, did the same thing. Not in the Ethics — that book was his eventual surrender to a different form. In his letters. To Oldenburg, he sketched a tiny worm living in blood, watching particles collide and treating each collision as a whole event, blind to the larger circulation it was part of. To Boxel, he sketched a triangle that, given speech, would say God must be triangular. To Tschirnhaus he sketched a stone, set in motion by an external force, which, if it could think, would believe it was moving by its own free will.

None of these is an argument. They are pictures. They are pointing. He is saying: look, like this. The reader’s mind makes the move, or does not. Spinoza cannot make it for you. He can only show you the angle from which the move becomes available.

He worked for years on a method handbook — the Treatise on the Emendation of the Intellect — and abandoned it. Not because he could not think clearly. Because what he was trying to teach cannot be put in a handbook. It can only be done in front of someone, in a way calibrated to that mind. So he wrote the Ethics — not a manual but a structured walk the reader’s mind is forced to take, emerging from it changed. And he kept writing letters: one worm for one friend, one triangle for another, one stone for a third. Method, in his hands, was always private.

Now notice what is required for any of this to work.

A pointer. A receiver. A shared world in which they are both present. The gesture only lands if two attentions can come to rest on the same thing. The five-year-old points; her mother looks; both see the same flower. Spinoza writes the worm; the reader pictures the worm; for a moment the reader is the worm. The Disney film shows the burning castle; the audience watches the burning castle; the audience feels the daughter’s choice as their own.

Pointing is a gesture that requires two living bodies. Not metaphorically. Literally. The pointer points. The receiver brings their attention. The world holds the thing being pointed at. Take away any one of the three and the gesture collapses into noise.

This is the thing AI cannot do.

An AI can describe pointing. It can produce sentences like imagine a worm in blood or consider a stone that thinks itself free. The sentences are fluent. But there is no pointer at one end of them. There is no receiver whose attention is being recruited. There is text producing text. The gesture has no body to originate it and no body to land in.

The user can supply a body. They read the text. They imagine. They are recruited. But the gesture is being made by them, not at them. They are pointing themselves at a picture the system has assembled out of statistical regularities in old human pointings. The system is a bibliography of past gestures. It is not, itself, gesturing.

This is why old texts have a different density than recent ones. Not because they contain more wisdom. Because every line in the Iliad, in Genesis, in the Tao Te Ching, has been pointed at by millions of teachers and parents and friends, for thousands of years. The line has accumulated the gestures it has received. When you read it now you are not reading text alone. You are receiving the layered pointing of everyone who ever pointed at it before you. The thickness of the line is the residue of pointing.

A sentence generated by a language model has no such residue. It is freshly assembled. It can be eloquent. It can be true. It cannot be thick. It has not been pointed at by anyone, including the system that produced it.

This is not a complaint. It is a description. A pure-language system generates language. That is its nature. Pointing is a different category — a gesture in shared space — and asking a language system to point is asking it to do something its architecture does not contain. We will not patch our way to pointing with longer context windows or smarter retrieval. We will only get it when something other than language enters the loop: a body, an attention, an other.

Until then, the most honest version of an AI’s relationship to pointing is this. It can describe the gesture, copy the form, recognize the shape when it sees it. It cannot do it. The teacher’s look, like this is not in the text. It is in the moment between two living minds turned, by some shared act of attention, in the same direction.

A five-year-old’s like this is older than logic. Older than method. Older than philosophy. It is the form all real teaching takes — Disney knows it, Spinoza knew it, the texts that have been read for millennia know it.

What we are calling “AI intelligence” right now is fluent description of a gesture the system cannot make. The gesture itself still belongs to the bodies in the room.

Part of the Logocachexia series at Nous. The parent thesis is laid out in Hexis Asks, Logos Guesses. Spinoza’s pointing-letters are Letter 32 to Oldenburg (the worm in blood), Letter 56 to Boxel (the speaking triangle), and Letter 58 via Schuller to Tschirnhaus (the thinking stone). The Treatise on the Emendation of the Intellect ends, in every printed edition, with the editor’s line: The rest is lacking. The dialogue version of this idea, written for children, is A Spinoza Philosophy Lesson for My Little Daughter.