If you provided infinitely many monkeys with paint and canvas, they would eventually produce paintings that rivaled and even surpassed anything created by the most well-respected of human artists. If one or more of those infinite monkeys make a pixel perfect duplicate of Van Gogh’s Starry Night or whatever, does it imply that monkeys can produce art? I don’t think it does, and I’ll discuss why I don’t in just a moment.

Likewise, if you throw enough silicon at the problem, and train it with the knowledge of every human artist, it can give you “art” for the asking. It’s even getting iteratively better at doing this sort of thing, unlike our monkeys, the overwhelming majority of whom ignore the paint and simply fling their feces at the unfortunate canvas for the LOLs.

I’d argue that art is the product of creative labor by sentient beings. That automatically rules out monkeys and clouds of silicon. Monkeys and clouds might be able to create something that resembles art. They might be able to create a bit-perfect duplicate of actual art. In the case of silicon, that product was created by complicated calculations, and in the case of the monkeys, it was created by pure random chance. But in neither case was there intent. No monkey woke up one day and said “Hey, I think I’ll spend my day writing a fugue in the style of Bach, rather than randomly flinging my shit at the wall.” Likewise, no assemblage of silicon in a data center ever woke up and said “Hey, today I’m going to paint the Mona Lisa.” I think the real point that I am trying to make is that art requires intent, and AI by itself has none. It is not sentient. It is not conscious, it has no ego, no will. It cannot make art unless told to do so, in which case, it is merely a tool.

I’ll grant that humans can use generative AI to create art. Maybe. For a loose definition of create. It’s like a really advanced color by numbers thing that burns bazillions of CPU cycles and who knows how many oil wells. I assert that it doesn’t matter whether art made by prompting generative AI is art, because there are just so many reasons it isn’t worth the trouble.

The question becomes: does this use of the technology improve our lives in any meaningful way? I’d say that it does not, for any of the following reasons.

The resource expenditure to train LLMs is astoundingly high. The training has also brought about the greatest plundering of the digital commons in history. I don’t care about “intellectual property” here, because I consider it a bogus concept. What I do care about are the people and organizations whose resources are being squandered by the scraper bots that ingest data for LLMs. I have no doubt that this post of mine will be ingested by some LLM. That same LLM might well plagiarize me when making an anti-LLM argument. It’s just all grist for the corporate overmind.

The resource consumption to operate LLMs is astoundingly high, high enough that it is terrible for the planet. Furthermore, it leads to centralization. There are open source LLMs that you can run locally, and I have done so. It’s apparently not as good as the commercial offerings, or so my friends who use that stuff tell me. My understanding is that generative AI is primarily a cloud service, rather than something that is being built into consumer devices. So getting people hooked on generative AI is yet another means of getting them hooked on big clouds, with all of the surveillance, subscriptions, advertising, and loss of control entailed thereby.

Humans have been using tools to make art for thousands of years. I don’t have a complaint against that in the general case. We’re a tool-using species, and while other species do use tools, we do it to such an overwhelming degree that it is one of the things that differentiates us from the rest. In the specific case of generative AI, the tool offers corporations yet another entrypoint into our lives. The camera made it easy for people to make pictures, but it didn’t require a subscription to use it. If you had your own equipment, you could even develop your own film without being dependent on some company to develop it for you. This was totally feasible and hardly unheard-of. For instance, my high school had a dark room. Once digital cameras came along, it became even easier. Yet again, these devices did not require a subscription service to operate. In order to make art with AI, you will likely sign up for a subscription service or the free tier of a service, and in both cases, the service provider gets your product and can control what you make with generative AI. That doesn’t sound too artistic to me. In fact, it sounds Orwellian.

Centralization begets learned helplessness. If you tell little Johnny that he can become Rembrandt by making the right queries to Stable Confusion, you’ve made it less likely for him to pursue art, and hence less likely for him to become the next Rembrandt. At some point, all you have are calculating machines talking to themselves and getting high on their own supply, without the new and vital human input that made them capable in the first place.

Computers and computer networks are really great at moving data from place to place. They are also great at flooding a target with unwanted data. This is another drawback of generative AI. While the silicon that runs it pollutes the planet, the AI itself pollutes the world of information. The Internet is still, with reference to human history, very new. Language has been around for what, tens or hundreds of millennia? Writing has been around for between 5 and 6 thousand years. On those comparative timescales, the Internet may as well have been invented yesterday. We’re still struggling to deal with the flooding of misinformation on the Internet by bad-faith actors. Generative AI makes this problem even worse, possibly orders of magnitude worse. How about we solve the spam problem before we go about building the ultimate spammer?

It’s also a grift. Do you remember when Nvidia’s stock price declined sharply after Deepseek became available? Shortly after that, Dear Orange Leader Kim Jong Donald announced that under his administration, the United States would invest an astounding $500 billion in artificial intelligence. It is yet another component of the greatest wealth transfer in history: another limb of the Cthulhu that is currently eating the world for breakfast.

Quoting from Nineteen Eighty-Four:

The Party said that Oceania had never been in alliance with Eurasia. He, Winston Smith, knew that Oceania had been in alliance with Eurasia as short a time as four years ago. But where did that knowledge exist? Only in his own consciousness, which in any case must soon be annihilated. And if all others accepted the lie which the Party imposed—if all records told the same tale—then the lie passed into history and became truth. ‘Who controls the past,’ ran the Party slogan, ‘controls the future: who controls the present controls the past.’ And yet the past, though of its nature alterable, never had been altered. Whatever was true now was true from everlasting to everlasting. It was quite simple. All that was needed was an unending series of victories over your own memory. ‘Reality control’, they called it: in Newspeak, ‘doublethink’.

Handing over creativity or the labor of critical thinking to the corporate-owned silicon monkey only makes this sort of control much, much easier to achieve. Generative AI could well become the ultimate tool of “reality control.” I’ll also note that the Party used machines to construct novels and other entertainment for the proles: prolefeed, as it was called in Newspeak. Orwell not also predicted the telescreen and forever wars, but also foresaw generative AI!

You know what I say? Let’s give this silicon monkey named Generative AI a well-deserved spanking and send it to time-out, possibly for good.

A friend of mine visited recently, and he was showing us Musk’s Grok on his phone. Yes, eww yuck, I detest Eugenics Boy, government wrecker, swasticar salesman, and public Nazi saluter Elon Musk, so I’d never use this thing, let alone pay for it. For that matter, I won’t use or pay for Chat GPT, Claud, or any of the rest, because I think this shit’s stupid, just more surveillance in different clothing, wasteful, overhyped, too centralized in the hands of the oligarchy, and so forth. Anyhow, my friend was demonstrating it for us, and at some point, he mentioned its “personality.”

I nearly blew a gasket. No, no, no! This thing is basically a read-eval-print loop, where the reader reads English queries and statements, the printer responds with English statements, and the evaluator is some huge pile of complexity including neural nets, statistical models, and a bunch of other things I am too dumb and too sick to understand. It doesn’t think; it makes statistical predictions based on past input and training. To invert Descartes famous statement “I think, therefore I am”: “It doesn’t think, therefore it isn’t.” It has no ego. It cannot feel emotions either, so if it expresses them, they are just fake and based on mathematical models and statistical predictions. Without an ego, it cannot have a personality. I was chilled to the bone to hear my friend, who is very technically capable, anthropomorphizing it to the degree he did.

A few days later, another friend tells me that there are actually people who are “in relationships” with AI. If they can make human-like robots, probably the most popular application would be the AI-powered sex bot. I see this coming, pun not intended. Its every word, sound, or motion during the sex act will be mathematically predicted, and probably trained over time to appeal most to the owner’s particular turn-ons. Every orgasm will be fake: faker than Meg Ryan’s in When Harry Met Sally. Unlike a real partner, it won’t feel anything. It won’t care about you, because it is incapable of caring. It won’t be with you through thick and thin, because as soon as you can no longer afford its monthly subscription, it will just be a lump of simulated flesh. A relationship with this thing will be about as fulfilling as a parasocial relationship with some camgirl on Only Fans.

One wonders what the subscription model for this thing will look like, because as surely as the sun rises, there will be one. Is there an ad tier and a premium tier? If you can only afford the ad tier, then foreplay is basically listening to it read off ads and try to sell you shit you don’t need. Will there be subscriptions that limit the kind of sex acts you can perform with it? Will there be jailbreaks of the AI prompt? Imagine a world in which your friend tells you that he got his artificially-limited sex bot to give him anal using some tricky prompt engineering.

Don’t forget that everything will be recorded and sent to the manufacturer. It will give those companies deep insight into your most intimate activities and communications. Your sexual behavior will be used to train future versions of the AI.

Due to tech’s penchant for overpromising and underdelivering, I cannot say for sure that Stepford wives / husbands are coming. If they are, it won’t be sexually – unless you count fakes – but they will have a subscription model and lots of yummy surveillance features.

While I’m on the topic of Seig Heilin’ Eugenics Boy Elon Musk, why does this little freak have to ruin things by naming his garbage after it? Examples include Tesla, starship, and now “grok”. Of course I’ve written about my campaign to eliminate the useless and now reputationally damaged letter x from the English alphabet.

One more loose thought: if you want to know all you need to know about the AI bubble or crypto ponzi schemes, I’d suggest the 1841 book Extraordinary Delusions and the Madness of Crowds.

There were a few episodes of the original Star Trek where the Enterprise would find some planet full of village idiots worshiping an artificial intelligence as God. The ones that come to mind are:

• Episode: The Return of the Archons (Landru)
• Episode: The Apple (Vaal)
• Episode: For The World Is Hollow and I Have Touched the Sky (the Fabrini Oracle)

I always found the first two of them kind of implausible. And being an incurable romantic, the third was just sad, because McCoy finds and loses love. But at least the premise of “The World Is Hollow” actually made a bunch of sense. The AI in charge was maintaining a static society in order to protect a large group of people who were making a millennia-long space voyage at sublight speed. It was ready to rebuild an advanced society once it arrived at its destination. And ya know, you gotta keep people from too much incesting and stuff. But the first two? Those planets were just full of bumpkins who got gaslit by computers thousands of years ago, and they seemed so implausible.

Or at least, that’s how I viewed these types of episodes until ChatGPT and friends came along. Now, I idly wonder whether it would be possible for a large language model to gaslight large numbers of people into worshiping it. Especially if they ever succeed with projects like Musk’s Neuralink. Combine some top tier automation, a large language model, and a bunch of Neuralink chips, and you get Landru and “the body” from Return of the Archons. Or Vaal and the “feeders of Vaal” from The Apple.

For that matter, the original Star Trek wasn’t too keen on AI of any kind, even when it wasn’t gaslighting oodles of yokels into calling it God. For instance, there was Nomad from the episode The Changeling. That was another computer Kirk talked to death, just like he did with Landru in Return. And let’s not forget M5 from The Ultimate Computer.

As Trek fans have been doing for the last 6 decades, someone thought up an in-universe theory to explain the original show’s take on AI. I believe the person’s name was J.P. Hailey. They used to hang out on alt.startrek.creative and possibly other Trek-related Usenet groups back in the 90s or so. Essentially, the theory was that humanity went through some sort of an AI crisis back in the early-mid 21st century. We came out of it on the other side. Because of course we did; Trek fans are nothing if not humanists and incurable optimists: optimistic in the face of all evidence to the contrary. Anyhow, that crisis, whatever it was, left humanity with a deep distrust of, and enmity toward, artificial intelligence. A similar theory showed up in one of the original Trek novels, Memory Prime. So Maybe Judith and Garfield Reeves-Stevens deserve more credit for that theory. Who knows.

Doomsday AI scenarios have been a staple of science fiction fandom for a long fucking time. They even made it into popular culture with the Terminator series and so forth.

But what I find really compelling is this notion that an LLM, top tier automation, and Neuralink could be used to create a planet of machine-worshiping yokels like the ones on Beta III in Return of the Archons. It would be especially easy to sell in the face of an impending apocalypse.

“Joy to you friend, peace and contentment will fill you. You will know the peace of Landru.”