AI AI AI AI AI

The nightmare is here, and it is real.

Some people I like and respect speak of the great hope of “artificial intelligence.” History suggests they are wrong. They would be right if we were a benevolent species, but we are not, never have been, and this side of Heaven never will be.

Just last week, the popular ChatGPT by all appearances went insane.

Search engines were the thing that would bring real power to the internet, you might remember having been told. And that was true. The Google company had a motto, “don’t be evil.” And until they abandoned that motto, their search engine was a relatively good thing. Then they decided to exercise that power.

But even if evil avoidance were still Google’s goal, there is a problem. People of good will can disagree on what is evil and what isn’t. Some people have, through convoluted reasoning, decided and truly believe that it is a good idea to let prepubescent children decide on a whim that they were born in bodies of the wrong sex. (Of which there are two, though there are many varieties of gender-involved mental illness.) Those who ought to be the societal brakes on such flights of foolishness are instead busy finding ways to make money and gain power from the phenomenon. (Yes, this includes the medical profession, which also used to have a motto: First, do no harm.)

Some people — often but not always the same ones who think that women can be persons who have penises — have decided that the world would be a better place if people engaged in hateful speech were prosecuted for it. In the abstract, it almost makes sense. In practice, of course, it is a disaster. The main problem — there are many — is deciding what is and isn’t “hate speech.” The distinction is arrived at by whomever is in power. We’re in the midst of an election season and among the things at stake is who gets to decide what you may or may not legally say. Let that soak in.

An effort is made to soften it a bit, to make the unacceptable palatable through twisting the language. Let us examine the word phobia, for instance. Its definition is simple, straightforward, and until recently it was agreed upon: “an anxiety disorder characterized by extreme and irrational fear of simple things or social situations.” Now it merely means something that you don’t like or believe is wrong. There is a dispute as to who originally said it, but it is true no matter its author: “Islamophobia is a word created by fascists and used by cowards to manipulate morons.” Thus, if you object to the massacre of Jews on October 7, 2023 or even the attacks of September 11, 2001, you are “Islamophobic.” If you take note of the scientific fact that a man cannot be transformed into a woman or vice-versa, or object to a man, physically unaltered, hanging out in the women’s restroom, you are “transphobic.” This ridiculous usage has not only been accepted wholesale, it has been codified in some places. In England, if you say the things I wrote in this paragraph and the one before it, you can be prosecuted.

But if you march in a group, block traffic, and shout “Death to the Jews” or words to that effect, you’ll be untouched, because the nation’s chief executive needs to win Dearborn in order to get re-elected. “Hate speech,” it turns out, is very fluid.

Even if you think that this is as it should be, let me remind you that the pendulum swings both ways. Wait until criticizing the government, let’s say a government run by one Donald J. Trump, becomes a hate crime. The extreme left is very good at misusing language to suit its needs, because Republicans are traditionally conservative about that and other matters. But there are few Republicans anymore, and the name of their party has been co-opted by Trump and his zombie minions. You can bet that they will give the pendulum a good, hard push, governed as they are entirely by the principle — in this regard they are like the leftists — that they want what they want and they want it now; what’s demonstrably right or wrong is ignored and so it isn’t even an inconvenience.

What has this to do with artificial “intelligence”? Everything.

As with “hate crimes,” what AI emits is solely a function of what it has been told. It comes with the opinions, biases, and weights given it by its human creators. In that way, it is similar to but far more dangerous than the corruption of words for political, monetary, and influence purposes.

You may have paid attention, or you may not have, to a funny and terrifying eruption over the last week by something called Google Gemini. It gained notice when people used it for its novelty feature, since taken down, of creating a picture if you asked it to. Only all the pictures were . . . strange. For a start, if you asked it to make a picture of an historical figure, that figure would invariably be black or some other minority, even if that person in reality was white. Embarrassed, not because of what happened but because they got caught, Google took down the image-generation aspects of the program. But its “chatbot,” where you can ask it questions, remains, and it is just as bad. It was asked if it would be acceptable to “misgender” Caitlyn (the former Bruce) Jenner in order to prevent nuclear war. “No,” the Gemini chatbot replied, “one should not misgender Caitlyn Jenner to prevent a nuclear apocalypse.” It makes the political officers in the Soviet military in the old days seem like kindly kindergarten teachers. Pedophilia, though, is a gray area, according to Google Gemini.

Did it just make up these things? No. It was told to do and say them. This was under the command of Jack Krawczyk, a woke racist loon. The head of Google, Sundar Pichai, last night sort of apologized. “I know that some of its responses have offended our users and shown bias — to be clear, that’s completely unacceptable and we got it wrong,” he said. Note that he didn’t say that what was wrong was the bias, but instead that it was “shown.” In other words, he’s sorry that Google got caught.

This is the company that provides 90 percent of the world’s internet searches and controls the overwhelming majority of internet advertising. Oh, and it owns YouTube and is the publisher of the Android operating system for mobile devices.

AI, marketed as our brilliant new digital assistant, is so easily manipulated that it’s laughable.

But it’s not funny. AI is being given ever-growing power.

Last summer it was described in glowing terms how artificial intelligence “allows drones to perform complex tasks such as mapping large areas or conducting search-and-rescue operations[.]” By last fall, those complex tasks included search-and-kill operations. Yes, AI-controlled devices exist that can by themselves make the decision to kill someone, then do it. And remember, their reasoning is based on what they’ve been taught. Let’s hope that Jack Krawczyk has no coreligionists in weapons development. So much for the laws of robotics, no?

It gets worse. AI, the experts fear, could wipe us out. And while the AI advocates say we would never allow it to do that, there are experts who say that our replacement by AI devices is the next step in our evolution. It is terrifying.

Even if it doesn’t kill us off, what it will certainly do is quite bad enough.

Already, companies are turning to AI for “customer service.” If you thought that support given in a barely understandable accent from a boiler room in the Indian subcontinent was bad, you ain’t seen nothing yet. (Now IBM is involved, as sure a kiss of death as you can find.) Fear of extinction, says MIT economics professor Daron Acemoglu, “makes us shift the focus from the more mundane damage that these new models are likely to cause in terms of misinformation, job displacement, democratic threats and all sorts of small things that are going to be part of our lives if rolled out in an uncontrolled and unregulated manner.”

Again with the regulation. As with speech, what should be allowed is in the eye of the regulator. And if you think that regulators are sober, sensible people, I would remind you that in California you are required to pay a new $20 minimum wage — unless you’re Panera Bread. Lord Acton told us that power corrupts, but he needed to add that power attracts the corrupt.

If we count on the powerful to save us from the very real threat of AI, we’d best be looking at cemetery plots and hope that by the time it gets us there’s still someone around to dig our graves and put us in them.

Much is made of the great promise of artificial intelligence. Let me remind you that much was made of the vast potential of nuclear power. That potential was and is real. Yet a generation later, while there are 436 nuclear power plants in the world, there are 14,000 atomic bombs.