Against AI in creative practice

I will never use AI in the production of text or in any other creative work. The only, tiniest exception is that I have been exposed to some AI on Google Search (I now block AI results).

I, like many other people, are frustrated by the intrusion of AI into work and creative fields, into apps, into the news, into our personal lives. It’s ubiquitous, or rather, talk about AI is ubiquitous.

In the ideology of tech billionaires, there is a turning away from the humanities: philosophy, religion, culture, art, human connectedness.

And a corresponding turning toward: the abstract, the numerical, the veneer of rationality and efficiency, outgrowing and detaching ourselves from our bodies, toward simple pleasure and the dissolution of the boundary between surface and depth, and fact and fiction. And, rising above everything else, is money.

Flattening

We all judge things superficially, to begin with. We start shallow, then get deeper, like wading into the sea. And before we leap, we judge the depth and constitution of the sea floor by looking at it from above — through the shallows. When ‘AI’ produces text, it flattens depth and shallowness. It’s good at producing copies, inexact copies, but superficially convincing copies. They pass our mind’s first tests — that yes, this seems legitimate. The surface is preserved, but the depths are either absent or twisted.

‘I’

The first person ‘speech’ of LLMs is confounding to some users. Often, an LLM prompt will be a question for which the answer was out there on the web. A few years ago, you could use a search engine to find a link to an article that answered your question. Now, the LLM will have read that article, and will regurgitate it back to you, using the first person pronoun to communicate. And the trick works, at least on some. The computer said ‘I’, so there must be something — someone — behind that ‘I’.

Epistemology

How do people ‘know’ things, in general? JTB — Justified True Belief (some people quibble that there’s an additional initial in that formulation, but let’s leave that aside for now.) JTB — the recipe for knowledge. If we consider an AI an agent capable of knowing things (unlikely, but for the sake of argument), the truth of the things it knows comes from the web and perhaps some basic inference or synthesis, and any justification will need to evidenced. Sometimes, it seems we treat intelligence as just a uni-dimensional vector — the more smarter you are, the more things you just know, as if by magic.

All the ‘smart’ stuff an AI does is off the back of human knowledge and understanding. All of it! When we say ‘wow’ to an AI, we should be saying ‘wow’ to ourselves.

Obsequiousness

The AIs ‘talk’ in a frankly bizarre and creepy manner. I do not want a computer to talk to me like this. I get squeamish when an automated email or an app is friendly to me, even before AI. I do not need my calculator to ask me how my day was.

Fanaticism

People are fanatical about AI, having abandoned patience or scepticism. Honestly, it’s very strange to see. People will boast that such-and-such industry is going, such-and-such thing is just round the corner, and there’s an expectation that this is an inevitable force of nature that we either ally with and totally accept, or we fall off the train and get left in the dirt.

The fact that there is such fanaticism raises a piercing scepticism in me. If it’s so revolutionary, so powerful, why not handle it with the great import it would surely necessitate. Even if AI is as special as its proponents say it is (and that’s an enormous ‘if’), then surely it’s still possible to bungle a rollout? If it is like a rocket that will shoot us to the stars, don’t we need to do all the necessary preparations so that it doesn’t explode on the launchpad? Where is the urgency coming from? Why?

Morality and value judgements

The condemned today. Things are bad, people are bad. Therefore, let’s make them worse. It seems people of all political persuasion are frustrated with the world today, and, good reason. But this unhappiness seems to generalise into an idea that the whole world is condemned — bad, fallen, broken — and therefore we deserve what’s coming for us, and why even try to make things better. Even pre-AI, a good chunk of politics wasn’t about improving things, it was about penalising your political enemy. It seems this is the pre-condition for a lot of AI hype: things are bad, we deserve what we get, fixing things is pointless, why bother, just love the decline. It’s also as if (and this is predominant in left wing circles) that ‘goodness’ is an incredibly lofty thing that we will never live up to, and if we think we are we should re-evaluate (which are both true in a certain sense) — but it leads to a striving toward a kind of impossible perfection. And since we won’t reach this lofty goal, we should suffer. Somehow, despite the superficial nihilism of everything, there’s a very basic morality coming through: the strong win, the weak suffer, trying to improve is futile, if not counterproductive.

Automation

If your job is automatable, even as an extremely poor simulation of your work, a poor facsimile, then your job deserves to go. Is this a sublimation of attitudes from the industrial revolution?

An LLM cannot produce a work of art. But it can produce a thousand extremely inferior simulacra of art. But to the AI lover, the fact that I can do a poor impersonation of you, well, that makes you the fool. Me! Outsmarting you, copying your voice and your mannerism, clumisly and exaggeratedly, mocking you, with no effort on my part. Fool on you for trying.

I’m not a doctor. But, a few years ago, I could take a surface level look at some text a doctor may have written, and make a good judgement as to the veracity of the content and the legitimacy of the author. I can trust my ability to pick a good doctor without becoming a doctor, at least to a degree and assuming a functioning web. And it’s not just the case for doctors — it applies to all kinds of professional or academic activities. It’s precisely this that is threatened by LLMs, because it can produce convincing — facile, obsequious, confident — but superficially convincing facsimiles of real work. We now will have to dig deeper to find the truth within.

This is true with authorship too (and by authorship, I mean all kinds of narrative creativity). Art is more opaque than academic or professional writing (in a sense) — its meanings are locked away, or entombed in something aesthetically captivating, like Tutenkhamun. Thus, an LLM could have an easier time producing a surface level approximation of art. Who wants this? And why? I want to read books by authors who love to write — they have good ideas, good insights, struggle day by day to get them on the page. I want to get inside another human’s head. It seems that some people envy the author, and thus try to undermine them.

God-gestation

Some people believe that AI is imminently going to disrupt the world. That superintelligent AI is nearly here. Never mind the fact that we won’t be able to control something more intelligent than ourselves (and the arrogance to think we could!) — we’ve got AI billionaires saying there’s a high probably that their product will cause doom to humanity. And gaping idiots lap this up. If you are in charge of building a nuclear reactor and you see early signs it’s about to meltdown, you stop and you solve the problem. You don’t keep going. It’s inherently paradoxical. These people think they’re going to create something they can’t control, and are boasting about it. But if they believed this, they’d stop? Like, obviously they’d stop? Or do they have a death drive? Or are they just fantastically large morons?

Further, there’s something very strange when the AI billionaire is sure your job is going (not his), despite the fact that he’s boasting he can’t do his own job properly (that is to say turn off the machine). The hubris is so thick it’s congealed into cheese, impossible to engage with sincerely because you’re holding your nose.

(The above point is true in the workplace, the boss saying to automate this or that, despite them not having the first idea what of how it is done in the first place.)

A common response is that these AI billionaires are simply lying. I assume they are lying, or deluded at the very least. But is that much better? They’re lying — causing mass panic and uncertainty — to meet some business goal. Don’t buy products from liars. But if the creators of these LLMs are so happy to lie, so ambivalent to the truth of their words and the meanings they connotate, what does that say for the very products they are making? Does the AI model lie because their creators are disinterested in truth, or are the creators lying because they’ve spent so long talking to their own creations they’ve abandoned sense and jumped head first into ego-stroking post-truth power fantasy?

Lies, prediction and hype

We rarely hear what AIs are doing. Rather, we hear what they could do. What they may do; how they will behave in 3 years time when fully developed and integrated. All is potential, what could be done, or may be done. We’ll hear ‘big picture’ dreams of work automation. But is this in fact happening? What makes ventures like this stumble is not found at the ideation stage, it’s along many of the small roadbumps on the way, like with the extremely slow rollout of self-driving cars.

Almost all people in the AI/tech space are making overly-confident guesses based on conversations with an obsequious interlocutor, happy to ignore any errors or hallucination as ‘well, think where we’ll be in six months’.

It’s as if ‘now’ is not important — the only thing that is important is an imminent future utopia, just around the corner, like heaven when we die.

We’ve had this with bitcoin (who is using that to pay for things) and we heard with with NFTs, and I’m feeling deju vu all over again.

Creeps

The AI billionaires are creeps, they are weirdos, what the hell are they doing?

It cannot be understated: these people are fucking weird. Marc Andreeson says he has no inner self, and that people of the past didn’t introspect. All he cares about is brute action. (Obviously the claims he made are preposterous, read some ancient philosophy or a history book!). Imagine trying to be a friend to someone like this.

This person, who is proudly barely a person in an emotional or intellectual sense, is shaping the technological, social and economic landscape, and he can’t or won’t turn his attention to his own motivations, his own meanings and beliefs. No wonder he likes AI: he practically is one himself.

What answers do we need?

Most of the big challenges facing societies and humanity are not related to text production. They are related to resource allocation, global warming and climate change, inequality and atomisation. LLMs help fix precisely none of these things, and often just make them worse.

This is not to mention the bubble, the imminent bubble that will wipe out the world economy.

Not to mention the AI data centres being built everywhere, which local residents are rightly furious about.

https://en.wikipedia.org/wiki/TESCREAL

https://www.youtube.com/@BetterOfflinePod

Don’t use AI in creative work. Don’t consume AI art. Be human.