The New Yorker has a new article about AI. On the one hand, it’s SkyNet, coming soon to a Terminator stalking you (with an endless supply of ammunition. There’s always an endless supply of ammunition. How is that? Who in the apocalypse is still forging bullets and making gunpowder? Who’s even mining the raw materials? Wouldn’t people go with swords? Much easier to maintain…). On the other hand it’s social media redux: probably not what you need, but benign except in how you use it. Cars, basically.
A passage, with commentary:
Artificial intelligence is a technical subject, but describing its future involves a literary truth: the stories we tell have shapes, and those shapes influence their content. There are always trade-offs. If you aim for reliable, levelheaded conservatism, you risk downplaying unlikely possibilities; if you bring imagination to bear, you might dwell on what’s interesting at the expense of what’s likely. Predictions can create an illusion of predictability that’s unwarranted in a fun-house world. In 2019, when I profiled the science-fiction novelist William Gibson, who is known for his prescience, he described a moment of panic: he’d thought he had a handle on the near future, he said, but “then I saw Trump coming down that escalator to announce his candidacy. All of my scenario modules went ‘beep-beep-beep.’ ” We were veering down an unexpected path.
Isaac Asimov wrote a fantasy trilogy that was accepted as science fiction because the premise was an imaginary science called “psychohistory” which has to be true because, as Asimov presented it, it was based on mathematics.
You know, the stuff that in the 19th century said bumblebees can’t fly and a train car moving faster than 30 mph would suck the air from the cars and kill the passengers? Okay, probably both old wives tales, but how about eugenics, the science that only became a pseudoscience after the Third Reich fell and we saw what the Nazis did with it?
Anyway Asimov invented psychohistory and gave it to Hari Selfon who predicted mathematically what was going to happen to humankind far into the future. Or at least to the end of the third book.
Mind you, I read the Foundation trilogy about 60 years ago (!), and I only remember enough about it to barely recognize the premise of the Apple TV+ show. And they didn’t improve it. Well, what’re you gonna do to the pig when all you have is lipstick? 💄
Anyway, what I do remember is the Mule, a figure who rose up to challenge and undermine Seldon’s plans for human salvation (IIRC, he predicted a hard time of 1000 years or so, and then a mathematically guided scientific utopia. Sound familiar?) The Mule was the “X” factor, the unpredictable unknown (the “unknown unknown “) which threatened to derail the best laid plans of mice and men and Hari Seldon.
Except, of course, science and math prevailed, because even the unknown unknown had been expected and was just part of the plan. Because science and mathematics can never fail, they can only be failed. And it’s an Asimov story, where science and mathematics are the truth which myths about deities never can be.
And this, in my childhood, is what passed for “intelligence.” Not in the entire world, mind. But in the world of a 10 year old boy besotted with science fiction, where people seemed to actually know things and made sense of the world. Well, until Harlan Ellison and J.G. Ballard came along. But that, as I too often say, is another story.
So Trump sent William Gibson down an “unexpected path”? Isn’t that the fundamental mistake of writers who think science fiction gives them a window on the future? Where is the “cyberspace” of Gibson’s imagination? As archaic and discarded as “Tron,” which tried to bring it to movies. Even serious dramas use computers as part of the “real world” they fictionalize, but could you sell Gibson’s cyberspace visuals as the equivalent of hacking, anymore. “Ghost in the Shell” anime did it pretty well, but does anybody really conceptualize the World Wide Web as a glittering wonderland of light shadow and dazzling devices with doors and ligh shows? Or even simply as a web?
Shit, we all know better. And how is AI working out lately?
Yup; that’s gonna take over the world any day now.
But the question is: what is “intelligence”? It can be data. Used that way it means information useful to governments, except it’s not taken to mean the information stored in the Library of Congress, maintained precisely to give Congress access to information it needs to govern the country (despite what the Roberts Court says, that responsibility is Congressional). So that’s a starting point, but not a very good one.
It usually means “smart people,” except who are they? I read somewhere that the drummer for the Police (I’ve already forgotten his name) was the son of a jazz musician. Seems Daddy was drafted during WWII, but they expected him to be worthy of being a grunt, because, you know…musician. But a bit of testing showed him to be a very intelligent person (because…musician) and they put him in military intelligence (the great oxymoron).
And don’t get me started on “the best and the brightest.” There are now recognized several kinds of intelligence. Some people are brilliant in highly technical fields, but hopeless in the simplest social situations. Others read people like books (my daughter, for one). Who is “more” intelligent?
And what kind of intelligence is AI? The intelligence of LBJ, the “Master of the Senate” and shepherd of some of the most impactful legislation in American history? Or the intelligence of Robert McNamara? Maybe the intelligence of Elon Musk? Or of Stanley Kubrick? Christopher Nolan? Vladimir Putin? Volodymyr Zelensky? Which of them has the “superior intelligence”?
Comic book villains always want to take over the world. I understand Marvel has a whole series going where that has happened. Why isn’t it on the screen yet? Maybe because the key to comic book movies is a certain amount of verisimilitude, and the idea of taking control of the entire world is just a bit too…Pinky and the Brain.
It just really can’t be done, or Europe would still be an imperial power over Asia and Africa and Russia and China and whatever parts the U.S. wasn’t still controlling. When we think about it for ten seconds the very idea is ludicrous because it simply isn’t physically possible. There are reasons Rome fell and the Khans couldn’t hold Asia and take Europe, or no one ruler ever dominated the African continent. But AI is going to take over the world…how, exactly?
In bad science fiction shows computers become dangerous because they can manipulate cables and wires and entrap our heroes (happened to Mulder confined in a trailer in “The X-Files”). And how a hard drive and a motherboard can make copper wire motive like the human brain can make muscle move bone is…never explained. Oh, yeah SkyNet can create Terminators in wholly automated factories supplied with raw materials by…what? Computerized mining and hauling machines at automated refineries that refine the ore and make the molds and smelt the material and blend the alloys and spin the wire and process the ore to create nuclear power plants (miniaturized) to run Arnold Schwarzenegger?
Well, when you put it that way…
There is a classic science fiction story, “With Folded Hands,” where robots built to serve humans do their job entirely too well, and without malice or brute force stop human beings from doing anything for themselves. It’s a trap the humans make and willingly walk into, while the robots do exactly what they understand they were designed to do. And the robots “understand” they are doing what humans want them to do, and making them happy. But in the end humans can make no decisions for themselves, can’t even make the robots stop, and so they must live and do nothing, with folded hands. That always struck me as the most realistic scenario of how humanity dooms itself with technology. We set the trap ourselves, and it springs on us by our own invention and acquiescence. Not by brute outside force and a “super intelligence” that regards us as bugs or fuel; but by creating the conditions that make us the victims of our own success as the machines we build as servants carry out their tasks in ways that prevent us from refusing their service.
Not with a bang, but a whimper.
I know AI is a defined term, but it’s defined within the field of computers, and often called “machine learning.” But what are the computer experts calling “learning”? Accumulation of data and how to manipulate it? Was Bach writing music? Or manipulating data? The two are not mutually exclusive, but “manipulating data” doesn’t necessarily produce the sublime work of Bach or Mozart or Beethoven. Does AI understand that? Do you? Bach understood music far better than I ever will. Einstein understood physics far better than I ever will. Why didn’t either of them take over the world?
And what is AI supposedly going to learn that they never learned? And why will we give it the artificial musculoskeletal structure to kill us all? And what would that look like?
What are machines learning, anyway? How to manipulate stock markets and financial markets? How to run mining operations? How to blast tunnels in the earth to extract copper? With what? The robots Elmo promised us he could make, until it turned out he couldn’t?
And is learning only the accumulation and manipulation of data? Is it not also understanding? Understanding not just what mathematics is (and what does AI do with Gödel’s theorem? Understand it? What would that mean?), but the philosophy of mathematics, of science, of human thought itself? It’s fairly clear animals think, but they don’t try to conquer us (theirs is a different, but not a lesser, intelligence). Why would AI necessarily be made in our greedy, self-centered image? Is intelligence ruthlessness? Or indifference? Or some mixture of both?
Is intelligence possible without emotion? Without bodily existence? We don’t know, we’ve never seen it that way. Arguably, it isn’t. So how would we know if a machine has intelligence?
And is that what it needs to take over the world? Why would it want to? Do we even want to?
