Fear & Coding:
An Idiot’s Guide to Artificial Intelligence
The smartest people I know who do personally work on AI think the scaremongering coming from people who don’t work on AI is lunacy. — Marc Andreessen, Twitter
“We made you as well as we could make you.”
“But not to last.”
— Blade Runner
Now that we live in an informational multiverse, it is possible to confront a field of knowledge completely new to you. I have recently been hired by a company to produce a philosophical appraisal of a number of connected subjects, one of these being Artificial Intelligence, or AI, a subject about which I know — or knew — next to nothing. Whereas in a past well within my lifetime this would have meant many book purchases and visits to libraries for the required literature, it is now of course possible to sit out in the courtyard with a laptop and browse the subject at my leisure, and mostly without charge or the inconvenience of travel.
This, of course, means that AI and its little solid-state sprites are already helping and guiding me. Our tacit opinion of AI is as mother’s little helper, whereas even brief exposure to contemporary theories of AI shows a fear of uncontrollable development and “gain of function” (to use a modish phrase) leading to the robot revolutions familiar to anyone who has seen or read Isaac Asimov’s I, Robot books and the resultant movie, Blade Runner or Philip K. Dick’s Do Androids Dream of Electric Sheep?, The Terminator, or Karel Čapek’s 1920 play Rossum’s Universal Robots.
The mythological element of inanimate life being given the human quality of thought and expression is an old one. The Greek myth of Pygmalion and Galatea has the former, a sculptor, creating the latter, a statue brought to life by Aphrodite, for whom she was sculpted in homage. The myth became a hit play, Pygmalion, by George Bernard Shaw, which was filmed more than once, including the wonderful musical version My Fair Lady, with Rex Harrison and Audrey Hepburn as Professor Henry Higgins and Eliza Doolittle, respectively. The Professor aims to win a bet that he can pass off a common flower-girl as a duchess by improving her speech. Curiously and rather brilliantly, the actress Wendy Hiller in the 1938 film version acts robotically once her accent has been brought up to the social mark. A precursor to the Turing test, perhaps.
The legend of the Golem, Mary Shelley’s Frankenstein, Kubrick’s 2001: A Space Odyssey — all of these are fearful myths, books, or films involving life being created where it ought not to be. A man who would ask too much of nature transgresses laws which should be sacrosanct. Of course, this is bred by the religious mind, but it still exists. Mortal transgression will last in literature for a long as mortals do.
And just as the creation of an intelligent being travels with the mythological baggage of mankind, so too do other myths — those of Prometheus, Faust, and their godchild Victor von Frankenstein, who declaims blasphemously, “It was the secrets of heaven and earth that I desired to learn.” And those secrets are made available due to what Frankenstein sees as a paradigm shift in scientific method:
A modern system of science had been introduced which possessed much greater powers than the ancient, because the powers of the latter were chimerical, while those of the former were real and practical.
Frankenstein was subtitled The Modern Prometheus.
Galatea was not real, nor was the Golem or Frankenstein’s monster or Roy Batty. But Deep Blue is, and so is Diana, Eliza the Chat-Bot, and armed drones that work out terrain as they encounter it. I armed myself with a few introductory videos and documentaries, and bought the first book on Amazon which seemed to me to be more a history of AI than tech-speak, which may as well be Aramaic to me. I chose Artificial Intelligence: A Guide for Thinking Humans by Melanie Mitchell. I think I chose well — or luckily — but you don’t have to be artificially intelligent to work out that if you read just one book on an unknown subject and pronounce it to be excellent, you fail to understand what throws a lot of the first modern artificially-intelligent entities: context.
Professor Mitchell is the Davis Professor of Complexity at the Santa Fe Institute, and so looks like a tough first date, but she provided me with a comfortable narrative history of AI without leading me too far into the dark forest of its lexicon. This piece could be considered a review of the book, but only from entry level.
Rather than take you down the rabbit warren (presumably inhabited and dug by robot rabbits with Einsteinian IQs), I will assume that if you are interested in and knowledgeable about AI, then you will already know this stuff. I read the book through and, as always, took copious notes, and reread it for a dominant theme. That was obvious all along: fear.
Professor Mitchell’s mentor was Douglas Hofstadter, who at least I had heard of. There are a few books I have read a lot about but never read, and one of them is Gödel, Escher, Bach: An Eternal Golden Braid by Hofstadter. It is something of a cult book and I have heard people speak of it with reverence. It sounds almost postmodern in the reviews in that it combines several ways of looking at the world, “all brought together to address the deep questions of intelligence, consciousness, and the sense of self-awareness that each human experiences so fundamentally can emerge from the non-intelligent, non-conscious substrate of biological cells.”
This started an itch I could not scratch throughout Professor Mitchell’s book, as enjoyable as it was. What is the monomaniacal drive that mandates AI to be a replica of its creators? Enter Frankenstein and Eldon Tyrell, the mission statement of whose company in Blade Runner is “More human than human.” I would be far more frightened of an AI that sits and thinks its thoughts in silence rather than bouncing around like big, stupid children and winning at chess. Professor Mitchell reports that Hofstadter was “terrified” of AI. I am not terrified of anything, really, but if I were it would be human beings — or a certain powerful caste among them — and not robot accountants who can use the parallel bars.
Deep Blue, the chess computer, can’t do anything else but play chess. EMI, the music program which fooled classical experts into thinking they were listening to works composed by Chopin when they were all a product of, as Asimov puts it through the mouth of Will Smith in I, Robot, “clockwork and lights,” would not survive long in a forest. Eliza the Chat-Bot wouldn’t know how to start a conversation spontaneously, or differently if she were attracted to the other conversant. People can do all those things. Why reinvent the wheel?
A faster calculator, if it is operative on large-scale computer systems, is doubtless extraordinary when reduced to the shorthand of bytes. But why is it a surprise that Deep Blue beat grandmaster Garry Kasparov and AlphaGo beat champion Go player Lee Sidol? Both chess and Go are gridded games with incontrovertible rules and therefore finite ramifications. It’s about speed of thought, particularly with those annoying clocks some chess players insist on using.
Now, as noted, this is my area of expertise in the same way as golf or molecular biology or Japanese are, but it was the fear factor that fascinated me from the start — and not just fear of the products of “machine learning” (a core concept of AI), but from some of its pioneers. Alan Turing, one of the great martyrs of AI due to his persecution for being homosexual, was the man who provided crucial work towards developing a machine which broke the Germans’ Enigma ciphers — the original of which I have seen in the Turing’s team’s former headquarters in Bletchley Park in England — and which gave the Allies a great advantage during the Second World War by allowing them to read the cipher codes that the Germans used for communications, and which they had believed were impossible to decode. Turing also said that “the statistical evidence . . . for telepathy is overwhelming.” Psychic robots I would definitely avoid.
Turing was not trying to create ersatz humans, just a machine that was an incredibly fast and infallible computer. “Computer,” incidentally, was originally applied to people who compute, generally women. Remember that during the next argument with your wife, should you have one. Computers were human, now they are robots, and AI — and Musk’s “transhumanism” — always seems to be aiming at a hybrid. Note that you are not asked by, for example, the CAPTCHA verification system to prove you are a human, just that you are “not a robot.” This leaves a space for a tertium quid lying perhaps somewhere between the two. I rather like that.
Perhaps hybrids walk among us already. Watch British Prime Minister (at least for this month) Rishi Sunak when he took office at 10 Downing Street. He waves at the surrounding press pack as you might expect him to, and his fixed rictus smile may just be social awkwardness. He doesn’t have Boris Johnson’s chummy slovenliness, Johnson always looking as though he was slightly drunk. But Sunak’s wave is extraordinary. Watch it here and tell me you don’t get a sense of binary ones and zeros rather than flesh and blood.
There are moments in Professor Mitchell’s informative and approachable book which will amuse the current readership, if not the Prof. herself, who thoroughly disapproves — like the time an associative visual program was asked to scan photographs and match images to the word “gorilla.” This bot had clearly never been to Wakanda, because it matched up a black family. When another program was given blocks depicting white and black faces, and was then asked to choose the most likely jailbirds, it went with national crime figures rather that social justice and chose the black faces.
Professor Mitchell is refreshingly skeptical about aspects of AI. Here, she describes the rather skewed habits of AI researchers:
A recurring recipe for AI research goes like this: Define a relatively simple, though useful, task and collect a large data set for testing machine performance on this task. Perform a limited measure of human ability on this data set. Set up a competition in which AI systems can vie to outperform one another on this data set until the human performance is met or exceeded.
The scrabble for research funding is, as you would expect, particularly fierce in the field of AI, and there is a sense that sometimes prizes are given to rather untalented pupils.
Now, as noted, I am approaching this like a little kid at his first sports event. I don’t really know what’s going on, but it is as tense as it is enjoyable. I am sure readers can point me to other works, and correct me in any errors.
Economist Sendhil Mullainthan puts the matter succinctly: “I am far more frightened of machine stupidity than machine intelligence.” But is something that goes against what humans think to be intelligent stupidity? Or is it something else? On a light note, if you haven’t seen the ‘70s Donald Cammell movie Demon Seed, this trailer contains the moral seedlings of a thorny briar patch, as well as being retro fun and starring Julie Christie.
Personally, I think there may be a reverse play in action here. The possibility of the AI bots which can play games, hold a conversation (there are many humans who haven’t learned how to do that), and order your groceries mutating into death-dealing titanium killers sounds like the purest sensationalism to me, but when I am found slain by a robot designed to vacuum the floor, that will serve me right.
But what is worth considering is the increasing tendency of the Left and the institutions that produce them to make humans think like machines. It may still be a case of “rubbish in, rubbish out” (certainly if you are reading Humanities subjects at a contemporary university), but these are the people who are and will be running things and countries, and they are being taught to repeat phrases like a player piano, or one of those dollies little girls used to have (boys too now, I guess) where a string was pulled and the doll said “Hello, Mama!” or whatever. Machine learning, to look at the current Left, is a two-way street.
In the end, I will give the last word to David Bowie, from the song “Savior Machine”:
President Joe once had a dream.
The world held his hand, gave their pledge
So he told them his scheme for a savior machine.
They called it the Prayer, its answer was law.
Its logic stopped stopped war, gave them food,
How they adored till it cried in its boredom.
Please don’t believe in me, please disagree with me.
Life is too easy, a plague seems quite feasible now.
Or maybe a war.
Or I may kill you all.
* * *
Counter-Currents has extended special privileges to those who donate $120 or more per year.
- First, donor comments will appear immediately instead of waiting in a moderation queue. (People who abuse this privilege will lose it.)
- Second, donors will have immediate access to all Counter-Currents posts. Non-donors will find that one post a day, five posts a week will be behind a “Paywall” and will be available to the general public after 30 days.
- Third, Paywall members have the ability to edit their comments.
- Fourth, Paywall members can “commission” a yearly article from Counter-Currents. Just send a question that you’d like to have discussed to [email protected]. (Obviously, the topics must be suitable to Counter-Currents and its broader project, as well as the interests and expertise of our writers.)
- Fifth, Paywall members will have access to the Counter-Currents Telegram group.
To get full access to all content behind the paywall, sign up here:
Paywall Gift Subscriptions
- your payment
- the recipient’s name
- the recipient’s email address
- your name
- your email address
To register, just fill out this form and we will walk you through the payment and registration process. There are a number of different payment options.