Cybernetic Machines: AI Art and Cultural Form

A “script” is a set of instructions fed into a cybernetic bio-technical machine called a “production company” that outputs a “movie”

A “composition” is a set of instructions fed into a cybernetic bio-technical machine called an “orchestra” that outputs a “symphony”.

A “blueprint” is a set of instructions fed into a cybernetic bio-technical machine called a “construction company” that outputs a “building”.

A “context model” is a set of instructions fed into a cybernetic bio-technical machine called an “AI” that outputs a “virtual world”.

Perhaps

Or perhaps all of the above.

These are all examples of “allographic arts” as introduced by Nelson Goodman back in 1962, versions of art that is crafted by others based on a set of instructions provided by the artist. this could be the director, the composer, the architect, as Goodman postulated, or a set of instructions followed by the Generative AI at the direction of the “Prompt Engineer”.

Of course “Prompt Engineer” is at once both too banal and too unrepresentative of what is going on in the artistic process here. The slightly more upscaled “Context Engineer” (for when one prompt isn’t enough) is similarly unsuitable here. Engineering has little to do with it at all, though much like our architect example above, engineering isn’t precluded from being a part of the process.

Perhaps it’s because the Generative AI tools are too new in their development to have a singular title, like composer or architect, or Madonna or Cher, and so we’re left with the dual names to describe them, by defining them as a variation on the thing that they are somewhat akin to. Think “software architect” or “3D modeler”. Too new not quite encapsulated in the name, the way “TV Producer” has collapsed into “showrunner” in the 21st century.

Maybe it’s in the name.

Or maybe it’s in what we make with it. The art form hasn’t coalesced yet. Again too new; too recently pulled from the primordial technocultural stew. In the early days of the form, we are left reproducing the elements of older media, the same way early television and film were often stage plays and vaudeville acts. We’re caught somewhere between Pong and Space Invaders in terms of development, with Elden Ring and GTA VI undreamed of in the distant horizon.

With that in mind, what will AI art actually look like? Once it comes into its own as cultural form? I hinter at it with Virtual Worlds above. These can be produced using traditional methods, of course, but maybe that’s but one way a fed set of prompts, of contexts, of world models can be realized. AI Art will almost assuredly look something barely glimpsed or imagined.

But I want to play in the holodeck for a moment.

Because I think that gets close to what we’re imagining here. The holodeck, famously introduced in the first episode of Star Trek: The Next Generation “Encounter At Farpoint” (airdate 1987-09-28) and subsequently retconned and chronologically re-situated as typical with enduring narratives, would allow for the cast and crew to input a set of commands into the computer and allow it to generate the setting, players, dialogue and the like, along a relatively broad range of possibilities. The computer onboard was massively powerful, and generated these holographic simulations with relative ease, but the show(s) always made that distinction between the computer of the ship, and the AI embodied in more ambulatory agents like Lieutenant Commander Data. It stands to reason that the computer of a faster-than-light starship some 250 years in the future would be more that capable at the task at hand.

So perhaps this is what we’re moving towards, where the cultural form of AI art is more akin to an “experience” crafted by an “Imagineer”, though perhaps not in a way akin to a theme park ride held under copyright by the Disney Corporation.

We’re getting closer.

Perhaps we don’t have the words yet because we don’t know what that cultural form will be. It’s had to tell from our Pong-centered viewpoint here.

So let’s try to re-work our formula from above:

A “prompt” is a set of instructions fed into a cybernetic bio-technical machine called an “AI” that outputs an “experience”.

Not bad, though perhaps a little generic. But what it gains in that genericity is that it is divorced from the digital. No “cyber” or “virtual” prefixes are to be found. And that allows for growth, for change, for possibility – for the cultural form of AI art to transcend the digital / material barrier, to allow for an full environment to be developed like within the holodeck, or for humans to interact with material AI agents, like the hosts within Westworld. We’re still bouncing around that “theme park” model, but there is art within that creation, of the building and shaping of a full sensory experience.

And the play is the thing, a phrase that was uttered in the holodeck on more than one occasion, I’m sure. So let’s leave it there, our recognition of the incipient cultural form of AI art, and go out into the world to hunt for new words, new worlds, and discover what the future might be.

Beyond where the robots go

I got out again this last weekend
Far enough away that there was no cell service, no internet
Far enough away that the smartphone might as well just stay locked in the glove compartment
Because it had no use aside from being a backup egg timer or flashlight

And it was good
I was able to touch all the grass
Out there beyond where the robots go

And after a few hours the din of social media faded
And I could no longer hear it chattering in my ears
And as I saw there in the stillness, reflecting in the dark
Much like now
I was left with a single question:

How do we reclaim the truth?

Because that’s what we’ve lost
And I don’t know if that’s the first step
Or the end goal

But I know that’s what we must do

The Mauve Pill

Been seeing a new type of post online, and there’s enough of them that it seems to be part of a trend. This might just be early days of it, but I thought I’d document my observations here, and return back to it as needed. I’m loosely calling it The Mauve Pill, for reasons we’ll get into in a bit.

Examples

Or this rather lengthy blog post:

https://eev.ee/blog/2025/07/03/the-rise-of-whatever

(There’s more, I’ll try and pull them out of the bookmarks shortly.)

What are the points in common of The Mauve Pill?

  • Anti-“content” – not as a style, but as the wholesale rejection of “content” as a meta-descriptor for various media, or thinking it is a new term rather one that has been in use since the 1960s (at least)
  • Anti-AI, either in general or rejecting that it can have good uses
  • Performative “left” politics, but without grounding or reflection (ie “against all”)
  • Uncritical assessment of new technology
  • Adoption of “en~ification” (which is not a new thing, really, just capitalism by another name)
  • A belief in the Dial-up Pastorale – not recognizing that the web was corporate from early on

There’s likely more, but these were some of the constellation of ideas I was observing.

Why Mauve?

Well, truth be told, I needed a color. We kinda ended up here by process of elimination.

Red and Blue are already taken, and Blue has connotations besides.

Red pill: These were people who, in the parlance of the community, had swallowed the “Matrix”-inspired “red pill” and seen “the truth.”
https://www.nytimes.com/2024/07/30/books/review/ellen-reeve-black-pill.html

Blue is staying asleep, stuck in the matrix.

However, maybe there’s another option?

I hear ya, Zizek. What other options do we have?

Black is taken too. Elle Reeve covers this in depth in her 2024 book of the same name, looking at “the black pill of nihilism” and how that has led to the pipelining of individuals to the alt-right movement.

So we need to find another option:

Silver? Would this stand for cyber, all shiny and chrome? Nah. Doesn’t seem to fit with what we’re observing.

Gold? Does this just mean (monetized or capitalized, or maybe just crypto)? In which case, no, this won’t work.

Rainbow? No, Pink, no? These have obvious associations which don’t apply here.

Green would be eco-friendly, perhaps? Brown might be too? Again, not sure either would apply in this instance.

Orange would be labour, perhaps? Or other worldwide movements, and as such it feels that this is taken.

So let’s go with Mauve. It kinda reminds me of old 4 color CRTs, visually close to the magenta provided by the CGA video standard, along with Cyan, White, and the black of the background.

Is it a problem?

A little, as it takes spaces in the discourse about dealing with current problems, but is as divorced from reality as some of those other “Pilled” movements.

There may be points where they (the authors of mauve-pilled content) identify an issue, but the pilling leads them to come to very odd conclusions, or looping in irrelevant examples in their train of thought on their posts.

And I notice that I may agree with some parts of the argument, or even the conclusion, but there’s enough fallacious, irrelevant, or specious reasoning in the logic chain that I feel that the end result is suspect.

So I gotta examine my own facts, knowledge and assumptions (which is fine, you always gotta check yourself), and then I end up questioning what I know?

Except, “I ain’t passed the bar, but I know a little bit”

It’s like: “I don’t know what you’re talking about, which means you (probably) don’t know what you’re talking about”

And it’s a problem because it’s “pilled” – posts along these lines get positive reinforcement from other pilled members of the community, and that leads them to think they are correct in their (flawed) analysis, further entrenching them in their idea and unable to be reasoned out of it by an expert in the area (f’rex me; see above – I know a little bit).

So because it occupies space, it closes off rational discourse about the subject, and we end up endlessly have to talk around whatever the Mauve-Pilled topic is online for the next decade or so.

There are literally bigger issues to deal with right now.

It reminds me of the Dial-Up Pastorale articles I was noticing last year, and I think many of them could fit within this trend. The DUP dialog has certainly continued, gaining steam and more attention as more people seek alternatives to large platforms. There is still a large amount of platform illiteracy involved there too – BlueSky, Surf/Newsmast and especially SubStack are not any better than the alternatives people are fleeing to them from; the switch just hasn’t been flipped yet to commodify the userbase. (Though the time may finally be coming for SubStack due to their current content policies.)

This commodification is one of the things they appear to share in common in another way: they all engage in what I like to call “Commodified Curation”, the provision a non-algorithmic internet experience. Curated, if you will. Right now they appear to be mostly benign, lying somewhat dormant within the social media ecosystems they’ve attached themselves to, but eventually the worm will turn, the switch will flip, and we’ll see the various stages of commodification and monetization take place once again.

We’ll return to both #commodifiedcuration and #platformilliteracy in the near future. For right now, the question is what to do about The Mauve Pilled?

I think for now it’s worth highlighting that it is a thing, and seeing if there are other examples. How much of it is just the zeitgeist, and how much of it is resistant to discussion. Much of what I’ve seen feels non-rational, like the other pilled groups, and as the saying goes: “you can’t reason someone out of a position they did not reason themselves into”. Give or take; it’s been a while since I’ve read Jonathan Swift.

If there’s going to be any engagement, it’ll likely start small. I’ll let you know how it goes here, and we’ll try and collect it for a future podcast episode.

The Ditch-Digger’s Dilemma

“Hey bud, I hear you’re a pretty good at digging ditches, do you want some work?”

“Sure thing. Whadaya need?”

“Just need you to dig this ditch the length of my property here alongside this rode.”

“Sure thing, lemme go grab my excavator, and we can get it done this afternoon.”

“Whoa, no, can’t have you using that excavator.”

“Why? Too loud? Bad hydraulics? What’s up?”

“Naw you gotta use this.” Holds up a spoon.

“Excuse me? That doesn’t make any sense. Lemme grab the excavator and we can get the job done in an a couple hours. A spoon will take forever.”

Yeah, well ya see, ya gotta use the spoon. It doesn’t count otherwise.

What? It doesn’t count?

Doesn’t count. Use the spoon, or don’t get paid.

This is the Ditch Digger’s Dilemma.

In an era of high technology, of power tools and machinery, of extensions of man, of solutions to the problems that can amplify our capabilities a thousand-fold, the labourer laments being forced to use the low power, backbreaking way in order for the work to count, for it be accepted.

And the reason why is frustrating, it’s maddening.

The reason? Aesthetics and ideology.

Which are choices, or course, but it means we can also make different choices. We’ll recognize that it can be difficult to step outside an ideological frame we’re in and see what other options there are, but they do exist.

When the Studio Ghibli transformer on huggingface.co was intially released, it didn’t garner that much attention, but when OpenAI released a similar tool in March of 2025, the profile of the platform suddenly had everyone talking about it. Most notably, was the creator of Studio Ghibli, Hayao Miyazaki, who called it “an insult to life itself“. The transformer (a type of AI deep learning model) allows the creation of sequences of text and elements of image quickly, in ways that are recognizable to a human audience, and in so doing eliminate a lot of the work in the creation of those images.

And the Studio Ghibli model of animation is very labor intensive – hand-drawn frames labored over for days, weeks, months. A famous 4 second clip for the 2013 film The Wind Rises took a single animator over 15 months to create.

Assembling all the scenes into a full film requires a large number of employees working for years to bring it to fruition. It can be laborious, exacting and backbreaking. But animation isn’t the first industry where we’ve seen the impact of this type of automation. Take a look at engineering in the 20th century.

On the left we see engineers and draftsmen prior to the introduction of CAD (Computer Aided Design) in the 1960s. On the right, traditional animation, albeit from a Banksy-created opening montage to the Simpsons (“MoneyBART”, S21E03, 2010). While the introduction of CAD and other digital tools has radically transformed what modern engineering shops look like, engineering is still a viable career path. The total number of engineers employed worldwide has only grown. They can use the digital tools to engage in more projects, more quickly, across a broader spectrum of fields than existed previously.

This is the crossroads art is standing at with the AI tools as well. Rather than have one man labour for 15 months on a single 4 seconds of film for someone else in someone else’s style, these creators can create, develop, and release their own stories, their own art, to a broader part of the population than before.

If art only counts if its creator suffered, then that is what you’re consuming – it’s part of the aesthetic. Or rather, the suffering outweighs the other aesthetic concerns. Aesthetic elements are secondary. And in an era of Late Capitalism, you’re condemning someone to suffer for money, in order to live.

Rejection of the AI tools by traditional and/or trained artists leads to some sub-par works being put up. Pipeline tools choose the first available option, or are only superficially curated, and then posted automatically using workflow automation tools like Zapier, make.com and n8n. It’s given us a lot of “slop”.

This slop is what is getting used to pejoratively describe most AI art on the web. Much of the use of the term traces back to an article from The Atlantic in August 2024, by Charlie Warzel, but I’ve managed to find instances of the use of the term #aislop on Mastodon going back to as early as October of 2023 (possibly) and January 2024 (definitely). The use of the term may have originated in other public spheres and social media prior to that however.

The slop is the easily generated content that’s now flooding the web as more people now find creative tools accessible to them, even if they lack the skill or training to truly make them shine. Which is a shame. With a little bit of work, some knowledge and training of art theory, you get to something more usable. You can move away from the realm of “slop”, and into something more expressive.

This ties into the shift within the AI realm from prompt engineering to context engineering (and glimpsed faintly on the horizon: “Worldbuilding” (but we’re going to talk about that in an episode real soon). Context engineering is “the practice of designing systems that decide what information an AI model sees before it generates a response” (Datacamp, 2025).

What is context engineering for a movie but the script? Or is there more to it? Hmmm. Perhaps…

Bastani, 2020

The tension with the AI tools is that at their core they are ultimately part of FALC – Fully Automated Luxury Communism (Bastani, 2020). The AI tools are a communistic technology – everything goes in (ref Soylent Culture), and ideally everyone can pull from them, but it is being pulled under tension by warring capitalists who seek to use it for their own ends. There are those in the tech sphere – the techno-capitalists (or cloudalists, vectoralists, “tech bros”; whatever philosophers are naming this group) building it and trying to capitalize on it in the usual ways: enforcing arbitrary tiers for uses, building hierarchy, installing friction, commodifying the users to sell to advertisers. You know, the usual. And they’re opposed on the other side by the rentiers seeking to extend their monopoly control on the IP, the intellectual property. They’re also capitalizing on the resource in the usual ways: gate-keeping access and distribution, locking in restrictive terms, installing friction via formats and region locking, commodifying the users to sell to advertisers. You know, the usual.

And like in any war, try not to get caught in the middle.

There is propaganda flying back and forth on both sides, of course, that you’re probably exposed to daily. This propaganda shapes and drives the discourse, leading to pilling and pipelines on both sides of the argument. With respect to AI, we’re starting to see a new pill arise, one that’s coloured Mauve.

Each of these groups of capitalists are facing challenges when dealing with the AI, and this much of this is inherent in the technology itself. For the techno-capitalists, trying to monetize it isn’t quite working – they’re losing a lot of money, and the traditional methods listed above aren’t recouping the investment. For the rentiers of the cultural industries, the paradox is that the AI tools do make the production of digital products easier, so they want to use them (at least at the C-level), but they’re facing pushback and resistance from the workers in those industries who view the tools (correctly) as a threat to their current employment.

So what’s a poor ditch digger to do? Pick sides in the ongoing War of Art? That’s one option, though the prospects if either side wins aren’t that great to be honest. Another would be STMOP. Rather, Seize the Means of Production. If the AI Tools are the new means of (digital) production, you need to grab that mop. And Sweep.

A Torrent of Starlings

A quick share of a recent story: creator Benn Jordan on detailed on their YouTube channel an experiment where they taught a starling the encoded waveform of a PNG image, and were then able to play it back. It’s a fascinating element in the field of ethology, (which we’ve touched on recently). Here’s the full video:

This has a lot in common with an old bit of internet lore: RFC 1149 (link), the original IP over Avian Carrier proposal, or “Carrier Pigeon Protocol”. These were originally developed as bit of a joke, using the format of the formal process for internet proposals (RFCs), but extending the protocol via looking at it as a proof-of-concept, to think about how it would actually work.

As the Wikipedia article mentions, there has been several attempts to make RFC 1149 work in the real world, to various degrees of success, but Benn Jordan’s example provides a novel approach, though with starlings instead of pigeons. (This has existed in science fiction for a while too of course, notably showing up with the fremen using bats as part of the distrans tech in Frank Herbert’s Dune from 1965.)

However, one of the issues with starlings is they lack the wayfinding capabilities of pigeons. So to effectively deliver a message, one might need to adopt a rather different approach. Enter starlings’ murmurations, the novel flocking ability where vast numbers of them shift and move through the sky in undulating waves.

A torrent of starlings, indeed.