AI Sucks: an examination


Narration
No AI was used to create this narration, just my own skills as an announcer and audio producer.
Music used: "Uncanny Plumbot" by T1na Badgraph1csghost.


AI Sucks, and if you disagree then You Suck Too.

I've had this image at the bottom of my homepage for about a year now, ever since Creative Commons came out in support of generative AI and I had to change all my licensing terms from CC-BY-NC-SA to All Rights Reserved. I used to acronymise "All Rights Reserved" to "ARR", which is always pronounced with a pirate brogue: "Arrgh!" As a composer, I hate using ARR because it turns off potential licensees and encourages piracy; but in my view, it was better than continuing to use CC and open my art up to the IP-distilling misinformation synthesisers. Well, almost all my art. To encourage other webmasters to reject generative AI in all its forms, I put this sign under a public-domain license. You can even see that right down there in small FreeSans, "Public Domain License".

Some people have accused me of kneejerk hate for emerging technologies for making this sign, and it would be easy at this point to get goaded into an old-style flame war; but the fact is that the corporate feudal state have dumped billions of dollars into a massive gaslight campaign, making people believe that AI is working. "Well, no, strictly speaking, the AI didn't come up with the right answer. But that's just because we need to refine the datasets a little more! Give us your data and we'll get right on fixing the problem!" The fact of the matter is, the technology is flawed at a fundamental level. Whatever GPT started out being, now it's just a morass of conflicting answers being churned out by a weighted-averages calculator in order to simulate artificial intelligence.

So, AI is not an "emerging technology" so much as a "get rich quick" scheme gone horribly awry. When OpenAI and others realised they had amassed enough data to compete with huge corporate data concerns, like Google and Facebook, they either sold out to the competition and retired while young or are going it alone with a board full of venture capitalists all demanding infinite profits immediately. I truly believe that the fall of NFT and the rise of AI are inextricably linked; the corporate feudal state was just beginning to adopt NFT as an infinite-growth currency when the whole thing imploded because Elon Musk couldn't shut his bloody gob, fired off a couple tweets, and made the entire enterprise implode almost overnight.

What do I mean by "corporate feudal state"? I use that term a lot, but what does it mean? As a system, corporate feudalism is exactly what it sounds like: the ruling class is occupied solely by billionaires with dollar signs where their brains should be, working the lower classes to actual death in factories and storehouses, while simultaneously manipulating world governments by installing insiders in decision-making positions thereby ensuring their "infinite growth" models can continue undisturbed by niggling little things like safety regulations. A corporate feudal state, therefore, is a government that has been corrupted in this manner. But I never say "a corporate feudal state", I always say "the corporate feudal state"; that's because you don't need more than one. Many such states would corrupt each other until there was only one, anyway; the one I'm talking about is the one controlled by Amazon, Google, Facebook, and Disney, with such figureheads as Elon Musk, Donald Trump, Bill Gates, and Jeff Bezos.

Anyway, as I was saying: when the collapse of NFT started making the corporations look stupid, they spent about a year scrabbling about looking for a new technology they could throw their fortunes into to prevent their stocks plummetting, and they found GPT3. OpenAI was a relatively unknown tech startup and DALL-E was being used mostly to make Tumblr memes. By suddenly adopting this new generation of AI, they could do several things all at once: they could save their reputations with the consumer base, they could convince their shareholders not to drop their shares like a math class and run away to the SEC, and they could use the underlying technology to downsize their human resources. Heretofore, this technology was being used chiefly to correlate data and make analyses; there was no such thing as "generative AI" until the corporate feudal state got involved.

Basically, "generative AI" doesn't exist. Computers are no more able to create their own original works now than they were back then; all that has been done is that these data correlators—essentially big weighted-averages calculators—have been told to correlate all the data on the entire internet and as many free databases as it can crawl. While this sounds impressive, consider how reliable the internet has traditionally been as a vessel for factual information. Cyberspace is the construct upon which the QAnon conspiracy theory was built; there are websites active right now that simultaneously deny the existence of other planets and convince people that aliens built the Giza pyramids, and espouse the idea that women are simultaneously superior to men and should exist to be subservient to them. When it comes to the actual content generation, the computer simply takes bits and pieces of the human-created works it's had dumped into its database and creates a Frankensteinian vivisection, using bits of this deviantArt user's work, bits of that Tumblr artist's work, bits of these 3 Artfight attacks, and arrives at less of an original work and more of stylistic plagiarism. While a human brain would be able to distill all of this information using a combination of data correlation, personal belief, and personal style, a computer is still not capable of that. That's why I called them "misinformation synthesisers" earlier: the weighted-averages machine is trying to find a middle ground with petabytes of contradictory data, failing to do so, and producing a result that makes no sense.

Then, of course, there's the environmental factor. I know you see phrases like that and immediately become Eric Cartman, but facts have the virtue of being true whether you believe in them or not. And the facts of this case are that generative AI wastes hundreds of gallons of water every single day. This is not to say that Google Datacentres haven't been doing the same thing for far longer, but ChatGPT uses almost 25 gallons of water per search. Consider also that ChatGPT is not the only generative AI process in active use; DALL-E, character.ai, Google, Midjourney, and about a dozen specialised fetish porn generators are also being used in tandem, to say nothing of the private AI research being done by defence contractors and other corporate feudal entities such as Amazon and Facebook. What? You thought they only used fluid-cooled processors in gaming computers? Well, they don't. Why did you think Google keeps building datacentres in so-called "developing countries" and in the impoverished areas of the United States? Since all the block-mining framework was still in place from cryptocurrency's heyday in 2021, the AI startups used their newfound wealth from their corporate investors to buy up those warehouses in South America and repurpose them for AI generation.

So, we arrive at the slogan: "AI Sucks and, if you disagree, then You Suck, Too". The intent here was less to instigate the Annoyance of Thought in the people who have already eaten the AI prolefeed and more to discourage others from falling down the same rabbithole. Generative AI doesn't work, and anyone who believes otherwise now belongs to a giant, multinational corporation. Step outside yourself for a moment and ask the question: "If AI is the trillion-dollar solution, what trillion-dollar problem is it solving?" The answer is "human creativity and innovation". Simply, generative AI exists so that no human ever has to be paid for a creative work ever again. Visual art, voice acting, music, certainly; but also creative or technical writing, slideshow layout, video creation, and even data entry. The point of this whole enterprise is to take humans out of skilled positions within the workforce and put them into a position where AI and robotics cannot be effective, such as customer service, manufacturing, and mining. More basically, grunt work. The kind of work that encourages consumerism at the end of the shift.

Don't misunderstand the slogan's underlying message. Generative is just one kind of AI amongst many. There's analytical AI, where the computer correlates a single, largely static set of data to come up with a necessarily accurate solution; gamers talk a lot about "enemy AI" or "NPC AI" in videogames, where the computer uses player inputs to simulate the appearance of a consciousness controlling the non-player characters; and there's general AI research, which has been going on basically since the rise of microcomputers in the early 1970s, researching whether it is possible to create an artificial intelligence of the same level as human intelligence. There's a time and place for all of these except generative AI, which, as we have seen, is not really AI and is incapable of generating anything without being able to rip off human creators.

--17 November 2024--
(Edited 7 May 2025)

HOME