The word computer first appeared in 1613. It was not the name of a machine. It was a job title. It referred to someone who calculated equations, someone who checked numbers alongside astronomers. Johannes Kepler employed computers. Three hundred years later, while designing a machine named after that word, Alan Turing defined it this way: "A human following fixed rules." Machines were born with human names from the beginning. But today's AI no longer follows fixed rules. It discovers rules on its own and sometimes creates new ones. Turing's definition is being completely overturned. So in an age when the rule-making entity has become a machine, what should humans be defined by?
Hampton, Virginia, 1943. The second floor of the west building at Langley Research Laboratory, a long room with a row of windows. The fluorescent light is even and white. Calculation sheets are piled on wooden desks. The clatter of Friden calculators alternates with the sound of pencil tips scraping across paper. Outside the windows, wind from the Chesapeake Bay ripples through the grass. The official job title of the women in this room was "computer."
In 1935, there were five women computers at Langley. When the war began, that number had grown to around 400 by 1946. Most of them were Black women. The West Area women earned half the salary of their male colleagues doing the same calculations. $1,440 a year. They solved the same equations, to the same degree of precision, in the same amount of time. Even during lunch, the cafeterias were segregated by race, so they could not leave the building. Some ate sandwiches at their desks while they calculated. Their names were hidden behind their job title, which later became the machine's.
Dorothy Vaughan, Mary Jackson, Katherine Johnson. They calculated rocket trajectories with pencils and calculators. In 1962, before John Glenn's Friendship 7 mission, IBM had calculated the reentry trajectory, but before launch he said: "If she says the numbers are good, I'm ready to go." The "she" he referred to was Katherine Johnson. He trusted not a machine, but a person who calculated like a machine.
In December 1959, the IBM 7090 arrived. The $2.9 million machine could not fit through the door, so the door had to be removed first. Once installation was complete, it took up an entire wall of the room. The temperature changed the moment it was switched on. Heat from tens of thousands of transistors computing simultaneously spread through the room, while cooling fans hummed low and steadily. Each time the magnetic tape reels turned, they added a clicking rhythm. The monthly rental alone was $63,500. That was dozens of times the annual salary of a Langley woman computer. In that room, two beings with the same name coexisted for the first time. The human computer and the machine computer. One of them would soon disappear.
Dorothy Vaughan knew it. She went to the library, borrowed D. D. McCracken's FORTRAN textbook, and read it at the kitchen table after her children had fallen asleep. FORTRAN. FORmula TRANslation. It meant translating formulas into language. What she learned was not calculation but translation. Moving the logic of one era into the grammar of the next. After mastering the textbook, she taught the entire team. When IBM had fully settled in, no one on her team was laid off. They all became programmers, and she became the supervisor of the programming section in the Analysis and Computation Division.
It was not resistance. Rather than break the newly emerged axe, she learned the axe's language first. Her choice still leaves us with a question. Was it possible because the language to be learned was visible, or was the essential thing the attitude of stepping forward to learn it first, even when it could not be seen?
Before long, "computer" became the name of a machine, and no one called people that anymore. Typewriter followed the same path. As machines became faster, the word crossed over to the machine's side, and people became typists, until even that faded away. Calculator and operator were the same. Language has a nature that constantly changes its owner toward whoever is most capable. Where did the people once referred to by those words go? They lived through a nameless period until new job titles appeared. How long that gap lasted is not well recorded.
In February 2026, Perplexity AI announced "Perplexity Computer." It is a system that orchestrates 19 AI models. When a user states a goal, it structures the detailed tasks and assigns the optimal model to each one. A word that moved from humans to machines has returned 400 years later in the form of a machine that directs the brains of machines. It resembles the way Dorothy Vaughan directed her team. Knowing who was good at which task, and assigning incoming work accordingly. The form changed, with a 60-year time lag.
Not long ago, Sam Altman said, "AGI has already swooshed past," while Dario Amodei said it had not arrived yet. They are building the same technology and looking at the same benchmarks, but they use completely different language about whether it is AGI. Hegel said that Minerva's owl begins its flight only as dusk falls. The truth of an era becomes visible only as that era is coming to an end. Their conflicting views are evidence that we are in the fog. Seen from outside, that fog would be a clear boundary. From inside, it is simply today's air.
It must have been the same on the night Dorothy Vaughan opened the FORTRAN textbook. She probably did not think she was part of the story of "the end of the human computer era." Her team simply needed to be able to keep working the following month. Historical turning points, from within them, are simply the things that need to be done today.
Today's AI is absorbing not human muscles but verbs. To write, to draw, to analyze. Those who lose their verbs are not given job titles. Those who lose "to translate" lose translator as well, and when "to diagnose" is lost, diagnostician begins to waver. The verb moves first, and the noun follows.
In 2024, Klarna replaced 700 customer support employees with AI. The title prompt engineer is already fading less than two years after appearing. In the 1940s, when one job title disappeared, another appeared. No matter whether the gap lasted a few years, it was filled. This time, no name is being given to the space. We still do not know whether titles are disappearing faster than new ones can appear, or whether this time there was never a space to fill in the first place. Earlier technological revolutions took away muscle and gave us new tools. Today's technology is learning human language itself.
Calculators are in museums. But the feeling of slowly moving the marker, of narrowing in on the answer with your fingertips, must still exist somewhere. Maybe it is the same feeling we have when we write prompts today. Even when the tools change, the core of what humans do has always been translation. From one language to another, from one era to the next.
The word "computer" lived for 400 years. From a human job title to the name of a machine, and then to the brand of an AI system. The word survived. Only the job title of the people it referred to disappeared. If, one day, AI fully acquires human language and creates new job titles of its own, who will those titles belong to?
I ask myself this question every day too. The title venture investor. Meeting people, listening to their stories, judging possibilities. Now that AI has begun analyzing market data, learning founders' patterns, and predicting returns, will this title one day be written as a single line in the functional specifications of some model? I still don't know. But I want to believe that simply carrying this question every day means I have already begun looking for the next language.
