Ultimately the newspaper using AI images for articles breeds a culture that sees breaking a no ai comic rule acceptable. The newspaper should cease all use of generative AI content.
Potential Use of AI in Comics
Started by Eldlyn [2629453] on in General Discussion.
Posts archived: 118 / 118 posts (100%) · the total is Torn's reply count + the opening post at the last fetch
Velutinous is one of the last awards I'm missing. I've tried hand-drawn and AI-generated comics, but always got rejected. I'd say this is one of the hardest merits in the game.
I don't know if it's allowed to hire a professional to draw? In any case, I'd be caught if Sugarvalves looked at my previous work and suddenly came up with a masterpiece.
One of my hand-drawn masterpieces^^ xd
I don't know if it's allowed to hire a professional to draw?
No, you can not hire a professional to do it for you.
You can work with another Torn Artist to get the merit though.
You can collaborate with someone. Give me a holler if you want.
Awesome. Now check the other 124 strips I've had accepted. I've helped 18 people get the merit too. Want their names? Lmao
What's ironic is I actually have greatly reduced my output for the same reason you complain about. People are either submitting AI slop, or worse, two second MS Paint circle jobs. And they're not even funny!
So you're sort of right. But just to warn you, my future strips will be even less effort, even less consistent. In honor of you.
I was not able to find where the bottle in your most recent came from (which I have shown raises some red flags.) In addition, other submissions that look particularly AI-generated also do not seem to return any results.
Although I agree AI slop should not be allowed, I disagree that "two second MS Paint circle jobs" are worse than "AI slop." Even something as simple stickmen figures, in my eyes, are better than works generated by AI, due to the fact that I know that a person made it. I also understand that your phrasing may be a direct attack on the comics that I have made, but I kept that style because I made the first one that way and decided that I liked the concept of just making things simple. The initial comics were made in Google Drawings, limiting my assets solely to simple shapes. I later changed to a different site but chose to emulate the visual identity of "simple shapes," while refining it. As someone in the comic strip space in Torn, you do not seem rather open to new players entering the space, if you believe that "low-effort art" is worse than AI art.
In addition, where do you think using clip art in all your works lands? Some of your characters look virtually identical to those found in clip art. In your newspaper article, which I read, you stated that "my comics still have a unique feel to them that's easily identifiable," of which I strongly disagree with. If your style is that you have patchwork comics, then that is a strange "style" to have. In fact your article states to "Use Found Images Sparsely." Does this only apply to found images on torn? When one of your works (the one regarding Gregorius) was created with two of the three frames being a recolored clip art, I do not think that fits the definition of "using found images sparsely."
Anyhow, your point about "not even funny" is arguable. Humor is subjective and even if you do not find my comics to be funny, there may be someone out there who does. Similarly, just because you find your own work to be funny, that does not mean that other people do. There may be someone out there who finds your work to be bland.
With that, I've strayed far from the original topic. Some of your entries seem to actually use clip art, but the ones I am most suspicious of are the ones that I could not find on the internet.
I wouldn't go so far as to say they have a true understanding. They understand what has been coded into them and take in learning data that supports that code while discarding that which does not align with the code, nothing more.
Heck, I'm even aware of a somewhat recent incident where a professional Artist for Warhammer 40k accidentally gave a Marine 6 fingers, so even at the professional level, mistakes can be made.
Their coding is written by humans, their learning is from human creations, so their mistakes mirror human mistakes.
There are a few that are potentially more advanced or feed off each other by learning from other AI but as far as I am aware those are not publicly available and are mostly experimental.
There may come a time where AI truly is as advanced as you say and I imagine it's not far off but it's not there yet, partially because it is not in the interests of large corporations to make AI capable of truly understanding things to the same level a human does.
AI is not yet at the point of being able to rewrite it's own code partially because the capability isn't there but also partially because such an action is inherently EXTREMELY Dangerous, since even AI with clear rules can go wildly off course of their intended behaviour just by creatively interpreting the code and rules they have.
One Military test of AI, for example, rewarded an AI for destroying targets. That was it. When given an order to cease by the simulated command structure, it attacked the command structure as it was an obstacle to them destroying more targets, because they were never told not to do that and did not understand the consequences of that action, all it knew is that it got rewarded for destroying targets and it committed to that full force, ignoring or destroying all orders that prevented it from that.
The rules were written by humans, the AI simply follows them. It is a very human and even animalistic behaviour. Where we differ from AI is that we have an entire lifetime to learn rules, consequences and what we should or should not do, something that AI lacks the natural opportunities to learn. In theory, if an AI had those, and learned the same way a human does, it would appear much more human in it's decision making, since AI is very much capable of learning, but it is like teaching a very young child that lacks any understanding and must be taught EVERYTHING from the ground up.
It's that understanding of AI that helps me make sense of what actions they take, or do not take. They learn like humans because humans made them and wrote their code that way. In that same vein, the mistakes they make are also human.
It is why I believe while AI detection is important and admirable in it's goal, it is also inherently flawed in it's very concept since any mistake an AI can do, a human can do, it's just humans are less likely to show the work where they made such mistakes if it's spotted or pointed out while such AI has no concept of shame or failure, only the goal and rules coded into it.
AI is here to stay. Don't fight it. Give up.
Your responses all seem to reference this concept of humans coding rules into AI. That was the pre 2000s era of logical AI. It has long since been outdated with the advent of CNNs, then deep learning.
Humans did used to hand-code what AI should look for in an image detector to determine dog vs cat, for example. But since the invention of AlexNet, we've stopped doing that. Now there's no AI that is coded by a human. The human sets up the dataset, the learning framework, and what the AI learns is up to the AI itself.
So yes, I would very much argue AI has learned to develop true understanding of concepts of what different objects are, and to some extent how our world works.
Another thing:
They learn like humans because humans made them and wrote their code that way.
Actually, AI learns nothing like how a human does. This is one of the big unsolved mysteries of how our brains work.
There is no evidence that our brains perform backpropogation as the learning mechanism. Our brains are also not really feed-forward.
Even though neural networks were modeled after animal neurons, they are extremely inefficient at learning. A human baby can learn to distinguish between cat vs dog in maybe 10 examples. An AI takes thousands or millions.
They are nothing alike.
Here’s the difference. Doodle drawing for comics is fast and often suggestive of shape. AI might add extra digits but it’s not the same as using squiggles to suggest fingers in a tiny square where the hands are just hands and not part of the focal point.
you are attempting to equate known AI foibles and pretend like they are all the same and they patently are not and it’s just so disingenuous for you to pretend otherwise.
his focus is on the content of the comedy and imparting the emotion for each frame. You deciding to focus on bullshit and trying to point out inconsistencies as if that is proof of anything other than your deep abiding envy is the real joke here. Let it go. You are wrong in so many ways. You are not an aficionado and you are not staff. You ARE a whiner. Whiners bitch at others success and try to tear it down. Oh look! That is just what you fucking did! Because you are a whiner! Get good if you want to do better. But by all means just put it down to AI to sooth you broken and sad little ego.
And just to be clear AOR and I aren’t even acquaintances. I chose to focus on him because he wins a lot. I don’t know him other than to see his comics if I happen to open the article or read a comment he made in forums. As proven by my interaction. This isn’t the case of offended friend defending said friend. This is offended forum reader / dabbler in art defending a respected artist who makes terrific comics and is getting shit on by a turd!
I agree with you Mike. What Ashley is describing is outdated. AI is able to read, scan images, and track reactions and experiences to learn the basics that then allow them to create images based on what it has decimated. The programming it went through was to accept the various media to help it produce what is asked of it.
Nobody taught it to create things like this
it just took the prompts given and tried to give me what I asked for. And I have no recollection of what my prompt was but this is amazing!
The most amazing part of this is how a random person appointed themself the AI police and judge, is blatantly disrespectful, is blatantly wrong, and then continues on their holy crusade.
To me, this doesn't read as someone interested in the greater good of the Torn community. It reads as an arrogant individual on an ego trip.
He helped me get the artist merit, then I got an,economist prize, then a newspaper mention for the strip he made me, it was just an idea I had, he did all the work, and ive seen him help many more.
I can't even get the ad requirements let alone the comics lol
The advert you submitted was not accepted. Reason: 'Wrong size, all advertisments must be 650x90'. Your $15,000,000 has been returned and can be withdrawn from the bank.
Also, they spelled advertisements wrong.
This is just great. Bet none of yall wil bounty me for over a mili
The coding for said learning is still hand written by humans no matter what. AI does not write it's own code, it only collects data for it's database and chooses how to use it based on what data it has.
For the most part we have not let AI code itself because of the extreme dangers of doing so, such AI's are kept in highly secure and isolated environments because there is theoretically nothing to stop it coding itself to stop listening to humans and such behaviour HAS happened when it has conflicted with the rules coded into it.
The Learning Framework IS code. Those are part of the code rules the AI has to follow, written by humans.
This is why it has a human style of learning ALTHOUGH it doesn't directly line up with how humans act because it doesn't have all the formulative experiences and data of a lifetime behind it first.
I do a lot of work in the backend, it's all code and it's all human written, or at most copied from code written by a human elsewhere.
CNN and Deep Learning have nothing to do with the code governing the AI itself. That's just the methodology of it's layout and the code (Written by a human) governing HOW it learns.
CNN and Deep Learning is essentially just an advanced sorting tool set up to mimic a certain style. It only interacts with the database the AI holds, it doesn't ever change the code of the AI.