A computer has never done anything but read and write. Pixels on a screen, a motor turning, money moving between accounts: underneath, every one is the same operation, ordered symbols in and ordered symbols out, transformed by rule. It was a literate technology before the phrase existed, fluent from the start in the languages we built for it to read.

That is the part the word “artificial intelligence” hides. The languages were ours, but we made them for the machine: precise, unambiguous, stripped of anything a compiler could not resolve. Learning them was work the human did. To use a computer, for most of its history, was to become literate in its tongues. We wrote in the machine’s grammar because the machine could not read ours.

So the literacy ran one direction for seventy years, from us toward it. Every programmer is someone who learned a second language to be understood by something that could not learn theirs. The burden of translation sat on the human, and our resentment of it had nowhere to go, because no other arrangement was available to build.

What moved across those decades was not the burden but the meeting point. The first languages were the machine’s own: binary, then assembly, a human thinking in registers and addresses, in the shape of the hardware. Everything after was us pulling the readable surface closer to how we actually think. C hid the registers. Python hid the memory. By a language like Ruby the code reads nearly like English, 3.times { feed_the_dog }, puts greeting unless asleep, and its designers called it exactly that: expressive, human.

But the human was always the one reaching, and a formal language can only come so close. It has to stay unambiguous or it will not run, so it can approach our language and never arrive. Languages like these are the near edge of that approach, about as human-shaped as a notation can be while a machine still needs us to make it exact. The gap between even those and the sentence you are reading is not a short distance on the same line. It is a boundary the whole trajectory pointed toward and could not cross from our side.

Consider what was on the screen that whole time. A page of text rendered the same on whatever machine opened the file, the order placed there once by a person and faithfully re-served, decade after decade. Coherent text on a screen meant a person had arranged it, somewhere upstream. Reliably enough that no one thought to doubt it. The machine carried the order. It did not author a syllable.

Then the math changed which way the literacy runs. Pass the accumulated record of human writing through enough of it and what concentrates is not a library of sentences but the structure of how we arrange them. A machine that holds that structure can read the one symbol system nobody built for it to read: ours. Not a formal language handed over for the purpose. The messy, accreted, correspondence-built thing that was only ever meant for us. Not the next step in the approach we had been making toward the machine. The approach completing from the other side.

That is the event. Not a mind approaching ours along the road this section already refused. A literacy, arriving from the other direction. For its whole history the machine waited for us to become literate in its languages. Now it has become literate in ours.

Literate is the exact word, and its precision includes the warning. To be literate is to read and write a language fluently. It is not to be right. A fluent reader of an inaccurate page produces fluent error, and the page looks no different for it. What sits on the other side is fluent in the record we deposited, which alone is no grip on what that record corresponds to. The fluency is real. It guarantees nothing about the truth of what it continues.

Every literacy we bother to name has the same parts. Financial, media, scientific: being literate in any of them means traction, the fluency to make the domain’s signs do something; discernment, telling a trustworthy sign from a mistaken or manipulative one, well enough for the stakes; and repair, noticing when a reading went wrong and recovering without handing the whole domain to someone else. Cynicism is not the advanced form. It is discernment’s autoimmune failure. What arrived from the other direction is the first part, at a scale no reader has had. The other two are what the meeting asks of us.

So the honest name for the thing is not a mind and not a threat. It is a literacy, running the other way for the first time. We spent seventy years learning to be understood by the machine. The new thing is only this: it can finally read us back.