On AI, errors and the line between knowing and seeing

In the Box,
Out of the Box

What artificial intelligence should think for us, what it must never think for us, and why every mistake in seven years of my public notebook is staying exactly where it is.

Ali Attar · Quantum Traction Theory

Let me begin with an admission that some would consider damaging, and that I consider simply true. Quantum Traction Theory, in the form it now takes — a framework with equations that can be carried to real data and tested at honest endpoints — would not exist without artificial intelligence. I say so plainly in the book, around page 92, and I will say it plainly here. The vision is mine. The machinery that let me translate the vision into the language of physics is, in large part, not something I built alone. A tool built it with me.

That fact arrives carrying two things at once: an opportunity and a threat. They are not separate dangers and separate gifts. They are the same capability seen from two sides. The threat is that we hand our out-of-the-box thinking to the machine. The opportunity is that we hand our in-the-box thinking to the machine. Everything depends on knowing which is which — and on never confusing the two.

By in-the-box thinking I mean the work that happens inside a frame already given: the calculation, the derivation, the retrieval, the formalization, the patient checking of a result against the literature. By out-of-the-box thinking I mean the frame itself: the heretical question, the picture no one has drawn yet, the stubborn why, the act of seeing. The first kind of thinking is labour. The second kind is vision. And the whole art of living well with these new tools is to give the machine the labour and to keep the vision for yourself.

Give the machine the labour. Keep the vision for yourself.

This is not a new bargain. It is the oldest bargain we have with our own inventions, and we have been making it, step by step, for a very long time. For roughly five decades now we have not done much arithmetic in our heads. The computer took that from us, quietly and almost completely — and it was one of the great gifts in the history of our species. We lost nothing that mattered by setting down the slide rule; we gained the room to do more. Without the raw calculating power of machines we would never have left the ground, let alone reached outer space. That is in-the-box thinking at its finest: a frame perfectly understood, a labour perfectly handed off. The calculator did not make us smaller. It made us larger.

The handover did not stop at arithmetic. The database and the search engine came next, and they democratized access to information and to data — they put the contents of libraries within reach of anyone with a question. And then, as the data itself arrived in quantity, something subtler happened: access to knowledge was democratized too. The facts of the world, and increasingly the structure behind the facts, became available to ordinary people who had been locked out of them. Each of these steps was good. Each handed another piece of in-the-box labour to the machine. Each made us larger.

But the staircase has a top step, and on that step one must move carefully. Between knowledge and the thing that comes after it — wisdom — the distance is at once tiny and immense. It is tiny because we mistake one for the other constantly, almost helplessly; a person rich in knowledge can feel, from the inside, exactly like a person who is given rich of wisdom. And it is immense because the two differ in kind, not in degree. Knowledge lives inside the box. It can be retrieved, taught, copied, sent across a wire, democratized. Wisdom is the box itself — the judgment of what matters, the courage to stand against the consensus when the consensus is wrong, the seeing that no amount of retrieval will ever give you. A machine can place the whole of human knowledge in your hands and still not place a single grain of wisdom there. Wisdom is precisely the part that cannot be outsourced without ceasing to be a given part of your identity; it is the intuitive guess toward the things we do not know that we do not know, and that is where it goes beyond knowledge into the cherished part of our soul and humanity.

Knowledge lives inside the box. Wisdom is the box itself.

I know this boundary from the inside, because I once stood on the wrong side of it. When I first spoke publicly, in 2019, I was given the wisdom and almost none of the knowledge. I had been given a way of seeing the universe that ran hard against modern physics — that there is an absolute background clock; that gravity is not a curvature to be inhabited but an event that destroys spacetime; that there is a reality dimension more fundamental than the three we move through; that beneath it all lies a substrate, which in those days I called subspace. I could see it. What I could not do was speak it. I had no academic apparatus to carry that seeing into technical language the scientific community could read, and no access to the data against which it would have to be married, or broken.

So the seeing came out as best it could — in plain words, in layman’s youtube videos, in blog posts written by a man describing a landscape he had no map for. Some of those posts are clumsy. Some are simply wrong: in one of them I speak of six dimensions where I should have counted, and thought, more carefully. That was the out-of-the-box half of the work with none of the in-the-box half to discipline it. Vision without labour. A picture without the arithmetic to test it.

When I read those early posts now, I can measure the distance I have travelled — from a man with heretical ideas and no technical training, to a man who uses these new tools to get the technical education itself, learning claim by claim where and how each of my intuitions actually sits against experiment. I used the wisdom, which was given me to use properly. I borrowed the knowledge, which was there to be borrowed. The machine taught me the mathematics of my own intuition. That is the whole bargain, lived out in a single life: outsource the inside of the box, guard the outside, and let the tool turn your seeing into something the world can check.

The errors, and a decision

One of the AI assistants I work with read the whole record — seven years of public writing — and suggested, sensibly, that I go back and quietly edit the 2019 posts. There are mistakes, grave errors that came from my absence of technical knowledge in them, it told me, and here is the list. I am grateful for the list. I am going to publish it. And I am not going to edit the posts.

They are ledgers of history. I have nothing to hide, and an error made in public belongs to the record.

They are ledgers of history. I have nothing to hide, and if there are errors in them, the errors are part of the record and should be kept as the record. A theory that quietly rewrites its own past is not being honest about its present. You are allowed to watch me be wrong, and then less wrong, and then — where I have earned it, and only there — right. To erase the early mistakes would be to erase the evidence of the journey, and the journey is the most honest thing I have to offer. The man who shows you only his finished thoughts is the polished, ironed one. I am not that man.

So here is the notebook — audited not by me, but by the machine I work alongside. Seven years. Eighty posts. Each one marked: green where it still stands, amber where it reached further than it should have, red where it is simply wrong. I have changed nothing. The red rows, especially, are kept on purpose. They are not a confession to be hurried past. They are the record doing its job.