When, exactly, did artificial intelligence get away from us?
Some may point to various moments: when it was commercialized; when the companies that depend on AI for their income decided to prioritize human-like interaction over utility; or perhaps farther back, when we decided to make the bots self-learning rather than teach them what they needed to know ourselves.
But Dr. Boli, who will be among those grumpy intellectual organic holdouts if the bots have not figured out a way to assassinate him, will suggest when the subject comes up that the process began with a series of decisions we made beginning in the 1950s—each one of which had the immediate effect of making some part of our interaction with computers easier, but the collective trend of which was to remove the computers from human control. By the time the large language models arrived, we had already made the decisions that would ultimately mean the bots had no need of us except as organic servants to keep the electricity flowing.
In the old days, when you wanted a computer to do something, you programmed it to do that thing and ran the program when it was needed. Programming in the very early days meant flipping switches or some such thing; then storing programs on a paper tape or punched cards eliminated the necessity for flipping all the switches correctly every time. But it was still necessary to enter the program one way or another each time you wanted the computer to do that thing.
Dr. Boli knew one of the programmers from the Whirlwind computer who came up with a revolutionary idea. It was often necessary to load a sequence of different programs to get the job done: first this one, then another one, and so on. Our programmer’s idea was that, when the computer had finished one program, it should be able to enter a kind of default state and load the next program by itself. He called it a “sequencing program,” but with hindsight we can see that, in a primitive form, he had come up with the idea of the operating system.
This was the first in the chain of decisions that brought us where we are today.
The point of the operating system is that, instead of human operators having to make the computer do every single thing it does, the computer itself should take care of the basic tasks. Thus the humans are relieved from repetitive drudgery. That’s what computers are for, after all.
But which tasks count as the basic ones? Which are the repetitive drudgery?
Today, if you look at your computer’s system monitor, or task manager, or whatever your operating system calls it, you will see that even when you have not asked the computer to do anything at all there is a constant flurry of invisible activity. Internet fora are full of people asking, “Why is my computer using all its disk capacity when I’m not running any apps at all?” All those people are Windows users, because Windows gradually eats itself up over the course of a year or so until it becomes nearly unusable. But the principle is the same with Linux or MacOS: the operating system is always doing things in the background because it might be convenient for you. You’ll thank me later, says the operating system.
Some of those things are little conveniences you never think about. We’re used to getting the correct time from our phones or computers, because they keep up with network time. Keeping the clock set correctly is a process that has to wake up and do its job every so often. It could be left up to the user to trigger the clock-setting program, but the user doesn’t want to be bothered. So the operating system does it all by itself.
Some of these conveniences go to extremes that cause considerable inconvenience. Windows is always loading programs into memory on the assumption that you might want to use them later on. It monitors which apps you use, and quietly shuffles them into memory so that, if you happen to decide to launch one of them, it will be ready right away. That works well on an expensive and powerful computer, but Windows is installed on the cheapest and most underpowered laptops, and they grind their puny drives forever trying to keep up with Windows’ guesses about what software you might need next.
And then, of course, there is the increasingly blurry line between what might be convenient for you and what might be profitable for the maker of the operating system. You never use FaceTime, you say? You don’t take pictures, so you have no need for Apple Photos? Well, that’s fine, but those apps aren’t going away. If you take them off your Macintosh, Apple will put them back with the next update. You may not need them, but Apple needs you to have them. Windows users find that Copilot worms its way into everything. Yes, you can uninstall it—but a few days later it’s back.
We have long since passed the point where the user decided what to run on a computer. Now the computer, obeying the whims of its capitalist overlords, makes most of those decisions. If you happen to want to run an image editor or a word processor, because that is the actual work you bought the computer for—well, go ahead, but you’re the computer’s lowest priority. The orders it got from Cupertino or Redmond come first.
The same applies to our phones, which are just small computers. Apple has long dictated, benevolently, what you are allowed to run on an iPhone. Now Google is closing Android, so that you will no longer be able to run software that has not been approved by Google—“approved” meaning that its makers have paid Google a fee and given Google their official government IDs, so that Google can report them to the proper authorities in China or Myanmar. What is telling is how few Android users that will affect: only a tiny number of hackers and enthusiasts install software outside Google’s control anyway, and cutting them off will have no perceivable effect on anyone’s profits.
We accept these things as the nature of our computers, because the things have crept up on us gradually. But collectively they mean that we, the users of phones and computers, have lost control. The phone or the computer will decide what we need, and it takes extraordinary effort (like finding a company that will sell and support a Linux phone, and paying $800 for it instead of getting it free with your phone plan) to wrest back any of that control from the machines.
Forty years ago, we decided what we wanted to run on a computer, and we ran it. Now we have to ask permission of the operating system, and we have to get in line behind the things the operating system thinks are more important. And these things had already happened before the arrival of artificial intelligence as a mass phenomenon.
So here is the answer to the question of when artificial intelligence got away from us. The answer is that our computers had already got away from us, and the bots simply stepped into the role that had been prepared for them by a series of decisions that began in about 1950.
So is there any good news for us, the organic life forms who are stuck in the bot-centered world?
Well, if you enjoy playing by the bots’ rules, then it’s all good news. The bots will give you shiny toys to play with; they will relieve you of the drudgery of making your own decisions, and they will reward you with sensory candy that will leave a sweet taste in your brain. Your cage will be beautifully gilded.
But there is another kind of good news for other kinds of people. As it becomes more impossible to live a digital life on your own terms, you may begin to discover that the digital life is not worth living. You can try taking a walk with your phone in your pocket instead of in your hand; you may discover beauty even in the bricks in the sidewalk or the weeds in a vacant lot. After a while, you might try the experiment of leaving the phone at home—cautiously at first, not straying too far from its supervision, but gradually taking more adventurous detours until you have managed a whole hour’s walk with no phone at all. You can begin to experience nature, which you have always loved as a series of AI-edited images in social media, as a real presence, with colors and textures and smells and sounds. You needn’t cut all the strings at once, of course: remind yourself that you can still validate the experience by tweeting about it later. But gradually, as the bots demonstrate more and more that they don’t need you, you may come to the conclusion that you don’t need them either.