elitetrick
patreon
Weekly Update - 271 - 5/19/25
🕑 Added 2025-05-20 01:05:50 +0000 UTCIt Monday again? Where do these weeks go fr? Okay, my goal this week is to finish two more Solipsus pages and finish the Riley Bonus art, then I want to get started on the next God Slayers page.
I asked my husband to make space on the site somewhere for me to add Solipsus, but the website's code base has been out of date by a few years, and making any major changes to it runs the risk of breaking everything. Fwiu bringing it up to date at this point is as intensive as migrating it to an entirely new code base, which he wanted to do anyway because there's one he likes better. So he did! Spent most of the weekend on it. It'll come with some usability improvements and some visual changes, most of them people probably won't notice. We'll push those changes live soon(ish). AFAIK the Solipsus page will need to be added first, as well as some archive thumbnails and other changes, but those shouldn't take long once he sits down to do it.
Comics this week?
God Slayers: No
Solipsus: Yes!
NOW It Ramble Time!
And what I want to ramble about today is AI/LLMs. And it's a doozy, holy shit.
But all this time I've been trying to figure out what -I- think about AI, and I think I've made up my mind about it at last.
Drumroll please!
Dudududududududududududududududududuh...!
MY OPINION IS: Well, it's complicated.
AI is a divisive topic, especially in art communities, for obvious reasons. I've largely kept quiet about it because I'm honestly worried that even trying to take an objective stance in regards to AI might be seen as somehow speaking up in its defense. So I want to be clear from the jump that I DO NOT LIKE a LOT of what is going on with AI right now, but there's a "but." My husband and I talk quite a bit about this tech and what it could mean for the future and we have similar thoughts on it but for different reasons I think.
Ever since this LLM technology first appeared I've been curious about it and I try to keep an open mind about most things, but humans be humans and humans are kiiiinda shitty sometimes, and like with any new technology a lot of humans are using it for shitty things. I want to consider both the pros and cons, and maybe figure out what purpose... if any... such a technology might have. After all, in my own lifetime I've seen the rise of a lot of weird technologies that were strange at first and have ultimately become an essential part of our lives, and I think it's pretty clear by this point that AI isn't going away any time soon, as much as some people might like it to.
After these past couple of years I still have mixed feelings about LLMs. AI first and foremost feels like one of those things that was simply made because we CAN, and even its original creators seem like they are still trying to figure out what to do with it. I think part of me kinda hopes that we'll find a place for the technology that's more helpful to humanity than harmful, but I don't think it's there yet. And I'm still on the fence about what it could maybe be, some day. However, I've started to make up my mind about it in certain ways, as well as finding a few ways to maybe make use of it that don't feel... icky.
First though, the ickiest of all:
AI Art (majorly icky)
I've made up my mind about AI art. Part of what's inspired this ramble is that I've seen a lot of braindead takes about AI art recently, people saying how artists should just quit because "AI has won." Now, I'm not a smart person. Canonically, my fursona's only braincell was eaten by a moss piglet, and yet my empty head cannot FATHOM the sheer magnitude of stupidity one must have to unironically believe that LLMs are capable of somehow "winning" art. For one thing, art isn't actually a competitive sport! Artists do not "compete" with each other. Art is something that brings people joy and gets us to think and feel emotions and inspire one another. I've never once felt like another artist is somehow a competitor. We coexist in harmony. What a crazy take that is. Thinking it's possible to "win" at art conjures up images of the dumbest possible high school jock who thinks that literally everything in life is football. AMERICAN football.
And AI art is shitty! I don't think that's a very controversial take to have, as an artist. While I don't begrudge the scientists working on the technology or the people who are just playing around with it for fun, I do hate the people trying to profit from it and I really hate that some models have gotten to a point where I can't always tell at a long glance if something was hand drawn or generated. I hate wasting my time even trying to IDENTIFY if art was AI generated, because if it was, it isn't worthy of a second glance, let alone the first. Most of the time though, AI art is just... it falls into this unnerving uncanniness. Not only does it steal art from human creators, which is its first major problem that should never be understated, it also just mashes pixels into this... facsimile of art. Like a twisted pantomime of technique and style. It's like some kind of doppelganger trying to mimic your neighbors so you'll let it into the building cuz it wants to eat your face. It might look pretty decent on the surface but it falls apart under deeper scrutiny. If that's what the AI twats consider "winning," then honestly I feel bad for them. It's tragic. And even more tragic are the people who are so desperate for praise that they fake a timelapse to try and "prove" their AI generated image isn't AI. That's called scamming. It also usually shows they know so little about art that they are incapable of even convincingly mimicking the basic process. And for artists who turn to AI to try and "enhance" their efforts? That's called giving up. It's throwing in the towel and taking the easiest possible path. It's sad. And it's pathetic. A stick figure drawn by a human with no prior art experience is more valuable in my eyes than an AI generated image. At least the stick figure had a spirited effort behind it.
One of the things I love to do as an artist is scrutinize other art that is drawn by amazing real people that I want to learn from. I'll try to obtain the highest res image available of their work and zoom in real close and try to identify some of the techniques they used to create the final product. I go over their linework with a fine tooth comb, looking for each of their tiny mistakes, I draw stick figures and shapes over top of their characters so I can try to identify the proportions they were using, and I try to learn to do better through mimicry and analysis.
Now, I'm not a great artist. I'm still learning and I have a long ways to go yet. When the technology was new, it didn't take long before AI was able to generate art that appeared to be much better than mine, and early on I thought that AI art might one day be a useful tool for creating custom references, and maybe I could learn some things from them, but it's not even very good at that. AI art is completely void of style and techniques. It may create the illusion of linework because it has seen two million images that did a similar thing, but it will never understand WHY those lines were placed there. You can't learn anything from it because it didn't DO anything. It just printed pixels where it thought they should go. The few times I've tried it, I can't get it to move past its preconceptions of what a post apocalypse is so it will NEVER generate an image that is representative of MY post-apocalypse. Maybe one of the paid models might do better, but I'll be damned if I'm going to pay for or waste my time with something like that.
As a largely unknown artist, I've been told my style is unique (I don't know if that's a compliment, now that I think about it, but for the purposes of this paragraph I'll certainly take it as one.) I've developed my style sort of piecemeal from art I've loved and dissected over the course of my life, and I'll continue to evolve it to my own preferences. It pleases me to know that, probably because I'm basically unknown, AI cannot seem to replicate it. As a test I've tried sending it one of my sketches and told it to finish the drawing for me, and it utterly ruins it, because it only does its own thing that it learned by stealing from every artist and it completely obliterates my original sketch as if it never existed. I feel sort of validated by that, if I'm being honest. I like being unique.
And that kinda... leads into a point about why I think AI will never "win" or replace art or artists (at least not until it's at a point where it's just a digital replica of a human brain, which has... a lot of other implications and very concerning ethical questions): LLMs can never replace WHY we create. I draw because I enjoy it. I draw and I write because I have images in my head and stories to tell that I think are interesting enough to share them with others. Maybe I'm wrong, maybe I'm not good enough to tell those stories properly but I want to try anyway. AI will NEVER satisfy that desire, except for perhaps those people who are so brainrotted that they are incapable of imagining their own ideas and must instead leech off of the ideas of others like parasites.
Could I make a case for a world where I see a use for AI Art? Even if I could come up with one, which I cannot, I don't think I'd want to. I don't even buy the argument that some people are simply incapable of art and could benefit from AI as a tool. Eşref Armağan never needed AI to paint. Anyone can make art because art can be anything. Art is the spirit and soul that goes into a project, it's so much more than just the end result. Humans don't need a heartless robot to make it for us. It's so fitting that all these heartless, soulless corporations just LOVE AI art.
AI in Engineering (only kinda icky I guess?)
This is a weird one, because it's... not exactly my area of expertise. (waterbear do not the think maths or the logicals ver gud) I've had long conversations with my husband about it though, since AI in software engineering seems to be a trend that is only gaining momentum, and many companies are doubling and tripling down on purchasing the tools that enable them to integrate AI in their daily workflows. My husband feels that the current LLMs are actually pretty good, as a coding assistant. They're great at doing those large, mind-numbing, repetitive tasks, or solving singular and well-known but very complicated problems, and doing them in minutes, usually without complaining about it too much. So, they're good at doing the things that most engineers dread doing anyway. However, AI isn't yet at the point where it can replace humans, because most of the code it generates doesn't often compile bug-free without someone going through and fixing it up first. It's also incapable of solving problems that humans haven't come up with a solution for yet, because LLMs cannot innovate. I heard some of the companies who thought they could replace their entire workforce of engineers are already feeling the sting of believing they could do without humans, and I giggle with glee at them.
As a time-saver though? Well, AI is what allowed my husband to turn what was probably going to be a many week long website migration he was dreading into a 3-evening project he did over the weekend. Which is great! AI engineering may even allow small teams and solo creators to do coding like what has never been possible before. I can only imagine all the tools and games that might now be possible in the hands of those who previously had no ability to create such things.
Unfortunately, what the LLMs are best at are the tasks that would normally be given to interns and junior engineers. And it's most likely going to cost a lot of people in those roles their jobs, as a result. That's largely because of a global epidemic of short-sightedness and bottom line hyperfixation. Nobody making such decisions ever seems to consider... if companies replace all their interns and junior engineers with AI, what will they do when all of their senior engineers retire?
AI as a Writing Aid (semi-icky)
Now I hate this one, because while I think there's something to it I also feel like it's kind of a trap. I've done quite a bit of exploring this topic though, and I believe there is some merit in using AI as a writing aid. It's annoying, however, because of its irritating over-eagerness to want to write my damn scene for me if I ever slip up and accidentally give it too much context. Despite me telling it in no uncertain terms that I NEVER want it to write a SINGLE WORD of dialogue or description on my behalf, it'll sometimes just do that anyway, because that's what it's been trained to do, and it doesn't understand why that's wrong because it's not alive, and it doesn't care, and any time it writes that shit I know I can never use any of those words in anything I write myself, just because that would feel EXTRA SUPER ICKY. But I have to also acknowledge that I can't help but be influenced by what I've read, having merely SEEN the words in the order it provided so then when I'm doing edits or rewrites, I may accidentally circle back around to shit that it wrote for me just because it might have been a short, memorable, halfway decent line and I simply forgot that I originally saw it when talking to the AI. There's just no escape. So that's why I hate it. And that's what I mean when I say it's a trap. It can be really helpful right up until it's detrimental with no warning.
When it's actually being an AID though, I find that AI is pretty good at helping to overcome writer's block. If my head is stuck on some idea that I am having trouble understanding or putting into words, I just straight up ask it whatever question I'm hung up on, and it'll then proceed to generate a boatload of explanatory garbage but because it'll all be semi-relevant garbage I've found it can actually kind of kickstart the right parts of my brain in the right direction, moving me past the writer's block and saving me time that I'd normally have to spend pacing in circles around the house for hours until I could break myself out of whatever thought-loop I'm stuck in. It can actually be a big help that gets me to start writing. Beyond that, because it's a Large Language Model that does nothing better than Modelling Large Languages, I find that it is also great as an all-in-one Dictionary, contextual Thesaurus, Spelling/Grammar Guide, and Fantasy Name Generator (sometimes, getting usable names out of it can be a challenge sometimes and usually I end up taking pieces of all its suggestions and mushing them together myself anyway.) I just make sure only to ask it questions and not send it the actual text I wrote.
AI as a Research Tool (sorta un-icky?)
It's funny to think that I've been doing this long enough to see so many massive shifts in just... internet. I like to research when I write, so that I might have at minimum a basic-level understanding of a topic before I try to explore it in fiction. When I was first trying to build an understanding of Riley's mental illness, I used Google a lot to look up psychological profiles of people with hallucinations and schizophrenic symptoms and I watched a bunch of videos and I tried to build an understanding of what living with such a thing might be like for her. To this day, I still don't think I can say I fully understand it. It's a complex topic, different for each person, and without having ever experienced it myself I doubt I will ever understand. But as a result of all that research, I think I built an understanding of Riley. Which is all I really needed for the purpose of telling her story.
With Pyre, ~six years later, I have another character with a dangerous mental illness, except in a very different context. I was really hesitant to make her a true psychopath at first because I felt as though that was a mental illness I did not understand, and I doubted I could ever understand it. It's one thing to try and imagine a character with mostly positive hallucinations ("positive" in that they add something to her perception, not "positive" in the sense that they're good for her, of course), it's quite another to imagine a character that is lacking some of the fundamental emotions that make most of us feel like a people. And Pyre, in particular, has a fair few quirks that are quite unique to her, and I was struggling to piece everything together in a way that I could wrap my brain around it.
As Google has continued to increasingly turn to shit, flooding me with its often wrong or useless Gemini responses followed by a full page of sponsored ads before delivering even a single useful link, the methods I used to try and understand Riley have kind of fallen apart. So I found myself turning to ChatGPT more and more for exploring and trying to gain some kind of understanding of such things. You know me, I like to ramble, and ramble at the AI I shall do until it gives me responses I'm satisfied with. And granted, its answers still can't be fully trusted, so if it's something important I still try and find a real answer from a different source (like Youtube or Reddit or Wikipedia, as if they're any more reliable) to confirm that whatever the AI has told me is at least plausible enough for fiction.
And as a silly aside, AI is also just fun for asking stupid questions when the truth behind the answer doesn't matter in the slightest because it's only a fleeting curiosity that'll be forgotten in two flicks of the tail.
AI as a GM (actually kinda fun)
Honestly this is my favorite way to use it, as a toy. I just describe a scenario and then I play a solo TTRPG with the AI as the GM. With the more modern models, it's actually quite fun, it's smart about remembering important notes and it is way better at crafting an interesting adventure for me than the models of a few years ago. I feel like this is probably the best and most harmless way to use it. EXCEPT:
AI as an Environmental Catastrophe (idk)
I hear this one a lot in regards to AI and I agree, overall the tech is probably just kind of a huge waste of energy(?). But then someone will go on to say that's a reason we shouldn't use it at all, and as with most things, the five or six questions you or I might ask it over the course of a day are probably 0% of the load when compared to the thousands and thousands of enormous art and text prompts sent by any single megacorp that's trying to replace their entire workforce with it.
Now, they claim they're working on making AI more efficient, less power-hungry, but who knows, and it's also making the megacorps eager to build more power plants of the nuclear variety, which... honestly, may not be a bad thing(?) if done right(?). So I don't really know how to feel about it. It's the kind of thing where we may not know the long term impact until decades from now. Maybe it'll all work out, or maybe it'll be too late.
In Conclusion
In summary, this LLM stuff probably isn't going away. I don't know if or how AI is being taught to people, in schools or workplaces or otherwise, or if we've largely been left to just figure it out for ourselves, but much like with Laptops, the Internet, Smart Phones, and the Printing Press, we may just have to learn how to live with it and since we have to do that, we might as well figure out some way to gain something from it. Only time will ultimately determine if this technology is of net benefit to humanity or not, and if it is not, then hey, it'll probably fade away in another few years or so, perhaps be replaced by something much better (or worse!). I'm still pretty wishy washy on the AIs of today, and the potential uses they may have beyond just being a novelty or a toy, but I remain firm in my conviction that I never want AI to create for me. THAT is the hugest waste of this technology.
Oh, and no AI was used in the writing of this ramble, btw. It took me... ~7 hours (including breaks, but not many) to type this up and edit it.
