Mr. Hands

I mean, I'm not going after any traditional procedural content, I just can't enjoy it in the same way I used to. I don't think it's wrong to make traditional procgen or that people should stop doing it or that others shouldn't enjoy it, but I have to acknowledge that my personal relationship to it has shifted in a really sad way. I'm very cognizant of how much human work goes into making a good procedural system!

This is why I don't use "AI" as a term, I use generative models or LLM or what have you, because AI is a label that does apply to fields adjacent to procgen and includes a ton of shit that isn't plagiarism, but the discourse has been so poisoned by the current grift's preferred use of language that no one can understand one another anymore.

I think a big difference among others is that procgen in games is typically very bespoke. There’s not really that i know of many highly generalized traditional procgen tools/systems because the way every game uses it is gonna be unique, so you’re gonna be building your own system (or adapting more granular tools) into whatever game you’re making; whereas using a gpt backend or whatever is something that is highly generalized that just kinda gets dropped in. Part of the idea is that you don’t have to build a special system anymore, you just plug in prompts to a cloud architecture you’ll never control. And if GPT tools go through a bad time and stop working as well, it’s also gonna make your game suck more. If the service changes or goes away or becomes more expensive suddenly your game might be fucked or just stop working. It turns perfectly good single player experiences into yet another live service game of sorts.

To build off the first paragraph (because I don't have much to add to the second beyond "I agree with it", and that feels too insubstantial to me): the main thing that's frustrated me regarding AI discourse is a lot of the good/necessary points against AI are muddled up with more reactionary stances; or at least ones that aren't thought out as much as they need to be. In particular, people will tend to reify intent in a way that aligns uncomfortably well with Walter Benajmin's negative theology of pure art. (Edit: I found the other Benjamin passage I originally wanted to quote. Scroll down to Louis-Philippe or the Interior.)

I suspect most of that comes down to who populates my social circles: artists, writers, musicians, game developers - individual artisans who rationally recognize this new technology as a threat to their livelihoods, but who lack the depth of perspective needed to mount a meaningful offense against it. (Or something. I'm surprised I was able to wrangle out something vaguely coherent.)

Reminds me of the headlines about Broken Sword containing AI art, when it was just software interpolation and upscaling of hand-drawn art.

As a longtime procgen dev and enjoyer, it's been a weird experience to look at present-day GenAI talking points, because they were all settled long ago in the space of Proc Gen.

• Proc Gen is a cost saving measure. Would Notch, Redigit, Derek Yu, and a whole slew of last decade's indies been able to punch above their weight and challenge the AAA's without it?
• Proc Gen can be soulless. But only if your designer decision was to push a button to get back a perlin heightmap, or Binary Space Partition and call it a day. Bad (or just old) roguelikes used to do that. Now, there is so much that an actual designer would do on top of it.
• Proc Gen is unintentional, that's the point- to see new behaviors emerge that weren't programmed in and surprise us in unexpected ways.

We've even had more experimental indies play with text-generating markov chains that were trained on centuries-old public domain text. Or level-generating WaveFunctionCollapse trained off of internally developed level presets.

All of this was an open and shut case in the last decade, where the advantages were clearly recognized, the flaws were acknowledged constructively, and the quirks were taken by the crazier artists to go into their own wild directions. The discussion was healthy.

If anything, those similarities between last decade's proc gen and this decade's GenAI should be a guiding light for how this new technology is supposed to be applied.

• Don't use it to empower AAA's to cut their teams of 10 designers to 1. Use it to empower indies where one person can now generate a world it would've taken 10 designers a decade to make by hand.
• Don't settle for just pressing a button with a one sentence prompt and call it a day. Give more nuanced details, train a fine-tune/LoRA with work you intentionally selected, and weave the results together with your own handcrafted content to make something that shows your specific style.
• Don't look at the quirks and hallucinations and declare it unfit for your use case. Take what you find as a novelty, and build your use case around it. Maybe it's useful, maybe it's not- but it's that exploration that constitutes what art is itself.

The only thing that's changed between last decade's markov experiments and this decade's GPT was that people started training models off things that weren't theirs. The answer is simple: don't. Developers have done it before, and they continue to do so. They provide the correct example.

Decades ago, designing a dungeon layout with a couple of setpieces and enemies used to be considered creative. We broke down all the techniques that made "good layouts" entertaining, and packaged them to run into infinity, for anyone else to use. If it were to be used by someone else, that program doesn't reflect the user's intentionality- there's no understanding of the nuances that went into it. But that's the nature of middleware- they'll use it to make something of their own. The automation didn't destroy the art of creating. We, the humans, just got more creative.

yeah @mtrc hit the nail on the head when he said so much of the evaluation of these programs is just vibes-based

my very tinfoil hat opinion has always been that ML criticism is always bound to turn around on real artists. and in the same way hypervigilant policing of images for "ai tells" leads to witch hunting an illustrator who's bad at drawing hands, i think the focus on "lack of author" as a primary problem with MLgen makes a turn on all procedural art inevitable.