the more i think about it, the only thing i really see AI doing is giving more power to the anti-social types and owning class.
for 99% of human history, a persons value was in their character. What they do daily, and this is summarized as reputation and known locally. Reputation has to be maintained daily - you can't accrue billions of reputations and then hold that as leverage over others.
industrialization killed the age of character and now the only means of leverage a worker has is what they know. They have to try and learn something valuable and rare so they can gain enough leverage to be worth keeping around by the owning class.
if work is gone, the only thing i see being left is what effect a person can have on how others feel. This will skew value even more to anti-social types and make life for working class even more volatile in an attention driven economy as the modern consumer has the attention span of a gnat.
Basically it's an attempt to monopolize the value of labor in a knowledge-based market - the ultimate power grab - because value of labor is really only thing people have in a system where means of production are owned by a few. Whatever the large scale effect will be, I doubt it will make life more stable, more just, more fair, or more enjoyable for the majority of people.
Well lets think about it a little more. I don't think a persons value is going to be reduced in the way your thinking should work be removed from the equation. Lets say we're in this dystopian future where AI has replaced the majority of human labor.
I think the key question to figure out what happens from there first is survival. By such a hypothetical future, UBI or some variant of it will be implemented because we'll still be in a monetary based society where it takes money to live. It'll be enough to afford the basics like some food, water, healthcare, etc ,but I would imagine thats about it. So what becomes of jobs if we're not forced to do them? Jobs would become optional at that point. They'd serve a few purposes in general I think:
-the ability to earn more than whatever UBI gives you (or perhaps UBI cuts out if you get a job?) so you can afford a better life -people who want to have purpose in their life -people who just want something to do instead of sitting around all day -people who have skills that AI/machines simply can't do that effectively
I consider those to be very realistic possibilities for what could happen should we find ourselves in that situation. Thats just focusing on people who want jobs however. As said by this point in our hypothetical scenario, UBI exists so there is nothing that really forces you to go to work. Its moreso a want /desire to work more than anything. Basically, this will create a society where human value is determined by those who want to get off their ass and actually be someone. I would say its better to argue the importance of a job is what would be reduced and as such, individualism/desire becomes more important.
Now would this be better than our current way of life....dunno.
From X (got on it by following links, I never liked twitter).
THIS is pure offensive insanity. Adobe is claiming royalty-free rights to copy and make new content out of customers' assets. Adobe is claiming that they can then sub-license customer assets to OTHER companies. They describe a "reasonable" use-case, but describe no limitations.
So we can take (scrape) whatever we want. I don't know about you but I don't want adobe to have any access to my work at all!
So not only can we take (scrape) your stuff we can also sell it! It's amazing these guys can walk, the b...s on them.
So they can take work from unannounced projects and pieces under NDA and sell/repurpose them at will? It's a bold strategy, Cotton. Let's see if it pays off for 'em.
And, mEtA announcing proudly that they are already training their stuff on anything that is posted "publicly" - full-on attempting the usual heist of "if it's posted on the pUbLiC iNtErNeT, then it's free to use".
why is cara any different than artstation or deviant art or anything else? If I can right click and save an image, surely the geniuses can still scrape it, right?
I bet somebody has a good answer, but I also bet that after some time it will come out that cara (or anything on the internet claiming to be some sort of safe haven) was just an opportunistic cash grab nothing burger, similar to green-washing. Say the words people want to hear, don't actually do anything. But maybe it is run by true purist who would die defending the dignity of artist... even if that is the case, I don't think you can protect your work on the internet from the richest thieves on the internet. I think the only way is to go full mujahideen - nothing digital whatsoever. Doubt that's feasible since artist at large compete with each other rather than cooperate, but probably better to just assume anything you post on the internet will always be scraped.
Oh, I am absolutely not saying that Cara is *the* solution - as a matter of fact I am not planning to join it anytime soon since I have my own self-hosted folio anyways. But I think it is well worth keeping an eye on, and the mere fact that 700 000 artists ditched IG for it in a matter of days says a lot about how quickly things can happen when people are fed up. The founder of Cara also has an excellent track record, and recently won her appeal in a case of plagiarism from a "photo-copier-painter" who stole her work, producing a painting based on it without a license.
About the platform, from what I've gathered : - Their stance is clear : no AI-vomit allowed. Meaning that at the very least, it could become a place for artists to browse art without having to think about AI-vomit, or having to double-check everything. - On scraping : the standout feature of Cara is the built-in Glaze support, which ultimately could make it the easiest way for artists to get their images Glazed, as not everyone can run the standalone version. Meaning that there are now three ways for people to Glaze their work : through Cara, through standalone Glaze, and through webGlaze (which requires submitting a request and art portfolio, hence similar to Cara in that regard.)
This is FUCKING INSANE!!! is this even legal? Now they're going to scrape my 3D models and textures and use them to train my replacement. Just fucking amazing...What am I supposed to make my textures in, GIMP??! Maybe I should start pirating their software.
Photopea is free, and an excellent replacement for Photoshop. I use it at home, it has nearly everything PS including PSD, hotkeys, 8bit/16bit/32bit, and a few extra filters PS doesn't have like a decent normal map filter and an excellent dilation filter.
Hadn't realised it's as old as dirt til I had a look, must be something in the water these vikings are drinking over there that speaks too longevity in the foss space....
OT - and in terms of 'ethical' so called business practices these tech entities are attempting to obfuscate of late, I'll just quote Lewis Hamilton (F1 racer) when covid initially hit the grand prix back in 2020, here in Melbourne simply as:
I got off the digital art train a few years back after growing bored/disinterested with it, but seeing as how things have changed since then with the rise of AI, I think I left it at the right time. Adobe and many other companies have lost it when it comes to scraping data for AI. It only reaffirms my belief that traditional art is going to make a comeback.
- Their stance is clear : no AI-vomit allowed. Meaning that at the very least, it could become a place for artists to browse art without having to think about AI-vomit, or having to double-check everything.
Cara doesn't allow AI... for now. They say their stance is against unethically sourced AI, but once the legal landscape of ethical and data privacy issues are cleared up with regulations, then they are open to allowing AI as long as it's clearly labeled. So, it's a welcomed reprieve from AI vomit, but only for the moment.
- Their stance is clear : no AI-vomit allowed. Meaning that at the very least, it could become a place for artists to browse art without having to think about AI-vomit, or having to double-check everything.
Cara doesn't allow AI... for now. They say their stance is against unethically sourced AI, but once the legal landscape of ethical and data privacy issues are cleared up with regulations, then they are open to allowing AI as long as it's clearly labeled. So, it's a welcomed reprieve from AI vomit, but only for the moment.
Such a thing can simply not exist with current gen tech tho? The amount of data needed to make these things "usable" is too damn high to ethically source the data. I guess it will come down to how ethical is defined... Looking at adobe, that term is very flexible.
Figma (temporarily) disables its AI due to copying Apple’s Weather App.
"the feature “uses off-the-shelf LLMs, combined with design systems we commissioned to be used by these models. The problem with this approach—which I outlined in my keynote last week—is that variability is too low.
We don’t know for sure because Figma hasn’t said, but this suggests that Apple’s weather app is somewhere in the training data, and that Make Design is replicating it too closely.
This stuff is fun to look at, but it always turns my stomach a bit, all that stolen content. Wish I could just be happy to play with these tools. The flurry of innovation is great to see, but at the core it's all rotten.
every time i've tried to use midjourney i leave with the same sensation as when i try to play certain types of games like vampire survivors as an example. If you don't know, vampire survivors is a slot machine with the facade of a game on top. It's an indie game that got super popular awhile back and most people remark about how addicting it is.
That's just one game and it is like a 1 to 1 mapping of a slot machine which makes sense because the guy who made it was a slot machine programmer as a day job before he made the game. But other games follow the same style to greater or lesser degree. And every time I played those games, after some time I am left with a feeling of profound disappointment in myself for having wasted time doing something so vapid and crassly manipulative.
It's the same feeling with the AI art. Every now and then I need to search google for some concepts, and if cant find something specific, why not try midjourney? You type in a prompt, you get back results. Not instant, but it takes just long enough you build some anticipation.... and then you get the result, and its allllllmost what you wanted. Just one more pull on the lever. The big one is coming...
Twenty minutes later I am thinking, "what the fuck I am doing with my time?" and then I go just try to do the thing myself, and i learn something new and it took ten minutes. And I get exactly what I wanted.
So i have only attempted to use it for concepting work but it feels as much counterproductive as it does productive to me.
Shit, that trailer is slick! Can't argue that the technology isn't amazing. Makes it hard to reconcile that "ick" feeling from ai art. But as easy it is to marvel at the technology, it's more a testament to human made art than the tech itself.
We were finally able to create a "Make Art" button...we just had to throw artists under the bus to do it. There really is ain't no such thing as a free lunch.
I want to detour for a second into using AI tools for non-art work- My husband who is one of the least techy people I know has starting using Microsoft's Copilot at his job. His work uses Teams for meetings so calls gets transcribed, and copilot does an impressively good job at doing things like summarizing the call, providing a list of follow up actions discussed on the call, identifying who people on the call are- what team they are on- what they specialize in. And these queries can return timestamps so they can be double checked. And it does a great job with heavy accents as well. For jobs were half your time is spent on meetings, it really does seem like a time saving tool.
This is the type of things I want to see AI used for, basically doing something a personal assistant would do.
But then again we've both seen people obviously using chatGPT putting out completely gibberish that might look good at first glance to people not familiar with the topic.
Yes, we know but: don't be that party-p***er! Anybody who's invested in AI stock : we don't want to hear the truth (yet).
Tbh, pretty obvious that hallucinating personal assistants and random image generators that are entirely dependant on the supplied (stolen) training data are not why you would want to set up a billion dollar data centre. They better find real world big money use cases for this stuff and right quick. Then again I have family members in big pharma and medical who assure me that we are well underway in making actual use of this stuff.
omg anime Neon Genesis Evangelion, Starcraft Terran Ghost, Mercy from Overwatch, Warframe, Sena Kashiwazaki from Boku wa Tomodachi ga Sukunai, Alita Battle Angel, full body, standing heroically in a industrial complex alleyway, masterful quality, best of Artstation, big titty waifu giantess material step on me omg I've been stepped on too hard, I'm dying, I'm dead now
I didn't know Adobe added AI images to their stock photography collection, it seems there is a filter you can use, but AI for stock is pretty interesting
Well in this case "interesting" means "giving the middle finger to Photographers who contributed years of their work to Stock ; and also destroying the platform itself" ...
Autodesk AI will also help you speed up rigging workflows in Maya. As any rigger or animator will tell you, characters with many complicated deformers running live, can quickly become cumbersome and slow to work with. Using AI, you’ll be able to approximate complex deformations on a character, and use that version for quick tasks like layout, blocking, and crowd work. Then, just switch back to the complex deformations to put the final polish on your animations.
"One such company is Wonder Dynamics, who have just released their Maya integration. Their tool animates, lights, and composes CG (Computer Graphics) characters into live action scenes automatically. We’ve worked with the Wonder team to enable our Maya users to seamlessly export CG characters from Autodesk Maya into Wonder Studio."
I think there are some big problems for Animators and Mocap guys. Even for the good ones out there, they can only do so much work being 1 person no matter how skilled they are. This new AI that Wonder Dynamics has will be able to do the work of an animator, lightning guy, rigger and character artist (in terms of blockout that is) all in one. Assuming the AI has competency at doing common animatable poses/actions (i.e. walking, jumping, talking), WD pretty much eliminates the need of 3-4 3D guys for any company or product that just needs basic 3D stuff you see in commercials or very simple games.
WD pretty much eliminates the need of 3-4 3D guys for any company or product that just needs basic 3D stuff you see in commercials or very simple games.
So guys....generally speaking, what should we do now to prepare for AI? This shit is getting better and better and I wonder in which direction I should develop my skillset to be somewhat prepared to what will happen to the industrie over the next let´s say 5 years. I mean let´s be real, the same thing that starts to happen now to concept artists for example will happen to 3D Artists too. Sure there will be some need for manual 3D work but it will become harder and harder to make a living out of this stuff.
@Seal40k personally I've been making my own game in my spare time, which is how I'm circumventing the issue. 3D art is one of the most complex and difficult jobs in game development, so if you know how to do that it isn't a big stretch to just do everything, especially with engines like UE and Unity around. However this might not be feasible/desirable for a lot of people, so developing new skills and branching out to other industries is always a good idea.
Lots of games come out dead on arrival, or are cancelled. You can be the best artist in the world, but if you're working with a studio with programmers and designers that don't know how to make an appealing game, or if you have a bad art lead/director, you're basically SoL. These people will end up using AI, and not only will their game suck, their art will too.
At SIGGRAPH this week, there are quite a few companies pushing the idea of 3d artists opting in, and these companies are promising to pay them based on the % their assets are used by consumers' prompts.
It all smells super fishy to me. I think right away of Hollywood accounting but also how difficult it is to discern just how much of any input data is actually used by an output.
Plus, the datasets will continually be diluted, both with other artists, but also with machine-created data.
At SIGGRAPH this week, there are quite a few companies pushing the idea of 3d artists opting in, and these companies are promising to pay them based on the % their assets are used by consumers' prompts.
I came here to respond to the original post but I think everyone is picking on Adobe for being Adobe, "Yes that quote of them reserve the rights to own your s*** if you choose to use their product has been there for a long time." I personally don't use Adobe and I thank the open source / people who were fed up with where Adobe and Autodesk were going and decided to invest heavily in tools like Affinity, CSP, Blender, etc.
As for AI, I mean we're talking about collection of shared data that somehow ended up with a stable diffusion, it is only as good as that shared data collection. As an indie game developer I have seen few things that AI can not do. Like, write a complex code and distribute it amongst multiple files, or create an image that exactly fits your imagination. These two factors will never be replaced but results from AI can always be used as references or some guide that may help you resolve your coders block or artist block, which had happened to me in the past.
In short, good tool, but extremely unreliable, just like the rest of the technologies.
Replies
Well lets think about it a little more. I don't think a persons value is going to be reduced in the way your thinking should work be removed from the equation. Lets say we're in this dystopian future where AI has replaced the majority of human labor.
I think the key question to figure out what happens from there first is survival. By such a hypothetical future, UBI or some variant of it will be implemented because we'll still be in a monetary based society where it takes money to live. It'll be enough to afford the basics like some food, water, healthcare, etc ,but I would imagine thats about it. So what becomes of jobs if we're not forced to do them? Jobs would become optional at that point. They'd serve a few purposes in general I think:
-the ability to earn more than whatever UBI gives you (or perhaps UBI cuts out if you get a job?) so you can afford a better life
-people who want to have purpose in their life
-people who just want something to do instead of sitting around all day
-people who have skills that AI/machines simply can't do that effectively
I consider those to be very realistic possibilities for what could happen should we find ourselves in that situation. Thats just focusing on people who want jobs however. As said by this point in our hypothetical scenario, UBI exists so there is nothing that really forces you to go to work. Its moreso a want /desire to work more than anything. Basically, this will create a society where human value is determined by those who want to get off their ass and actually be someone. I would say its better to argue the importance of a job is what would be reduced and as such, individualism/desire becomes more important.
Now would this be better than our current way of life....dunno.
It's a bold strategy, Cotton. Let's see if it pays off for 'em.
https://petapixel.com/2024/05/14/meta-is-using-your-instagram-photos-to-train-its-amazing-ai-image-generator/
This + the Adobe debacle + 700 000 people joining Cara overnight, all in the span of a few weeks. Yay tech world !
I bet somebody has a good answer, but I also bet that after some time it will come out that cara (or anything on the internet claiming to be some sort of safe haven) was just an opportunistic cash grab nothing burger, similar to green-washing. Say the words people want to hear, don't actually do anything.
But maybe it is run by true purist who would die defending the dignity of artist... even if that is the case, I don't think you can protect your work on the internet from the richest thieves on the internet. I think the only way is to go full mujahideen - nothing digital whatsoever. Doubt that's feasible since artist at large compete with each other rather than cooperate, but probably better to just assume anything you post on the internet will always be scraped.
https://techcrunch.com/2024/06/06/a-social-app-for-creatives-cara-grew-from-40k-to-650k-users-in-a-week-because-artists-are-fed-up-with-metas-ai-policies/
About the platform, from what I've gathered :
- Their stance is clear : no AI-vomit allowed. Meaning that at the very least, it could become a place for artists to browse art without having to think about AI-vomit, or having to double-check everything.
- On scraping : the standout feature of Cara is the built-in Glaze support, which ultimately could make it the easiest way for artists to get their images Glazed, as not everyone can run the standalone version. Meaning that there are now three ways for people to Glaze their work : through Cara, through standalone Glaze, and through webGlaze (which requires submitting a request and art portfolio, hence similar to Cara in that regard.)
Now they're going to scrape my 3D models and textures and use them to train my replacement. Just fucking amazing...What am I supposed to make my textures in, GIMP??!
Maybe I should start pirating their software.
+1 for Krita btw
Also +1 for Krita! In fact there are alternative open software solutions for most of the stuff people here need.
For what it's worth, the lead developer as far as I'm aware is Halla Rempt, and they're Dutch.
AI Hype is completely out of control - especially since ChatGPT-4o - YouTube
rundown of big lies starts at 12:37
https://survey.adobe.com/jfe/form/SV_4Hfb49upjuZCcl0
Rapidly Generate 3D Assets for Virtual Worlds with Generative AI
https://developer.nvidia.com/blog/rapidly-generate-3d-assets-for-virtual-worlds-with-generative-ai/Looks like nvidia is building a huge library of 3D content which it will use to train its Ai generation/creation models.
The amount of data needed to make these things "usable" is too damn high to ethically source the data.
I guess it will come down to how ethical is defined... Looking at adobe, that term is very flexible.
For 3D artists, there's some info here https://blogs.autodesk.com/media-and-entertainment/2023/11/13/autodesk-ai-media-and-entertainment/
That's just one game and it is like a 1 to 1 mapping of a slot machine which makes sense because the guy who made it was a slot machine programmer as a day job before he made the game. But other games follow the same style to greater or lesser degree. And every time I played those games, after some time I am left with a feeling of profound disappointment in myself for having wasted time doing something so vapid and crassly manipulative.
It's the same feeling with the AI art. Every now and then I need to search google for some concepts, and if cant find something specific, why not try midjourney? You type in a prompt, you get back results. Not instant, but it takes just long enough you build some anticipation.... and then you get the result, and its allllllmost what you wanted. Just one more pull on the lever. The big one is coming...
Twenty minutes later I am thinking, "what the fuck I am doing with my time?" and then I go just try to do the thing myself, and i learn something new and it took ten minutes. And I get exactly what I wanted.
So i have only attempted to use it for concepting work but it feels as much counterproductive as it does productive to me.
But as easy it is to marvel at the technology, it's more a testament to human made art than the tech itself.
We were finally able to create a "Make Art" button...we just had to throw artists under the bus to do it.
There really is ain't no such thing as a free lunch.
This is the type of things I want to see AI used for, basically doing something a personal assistant would do.
But then again we've both seen people obviously using chatGPT putting out completely gibberish that might look good at first glance to people not familiar with the topic.
omg anime Neon Genesis Evangelion, Starcraft Terran Ghost, Mercy from Overwatch, Warframe, Sena Kashiwazaki from Boku wa Tomodachi ga Sukunai, Alita Battle Angel, full body, standing heroically in a industrial complex alleyway, masterful quality, best of Artstation, big titty waifu giantess material step on me
I didn't know Adobe added AI images to their stock photography collection, it seems there is a filter you can use, but AI for stock is pretty interesting
This is such a weird era.
For 3D artists, there's some info here https://blogs.autodesk.com/media-and-entertainment/2023/11/13/autodesk-ai-media-and-entertainment/
Seems interesting although when I read this,
I think there are some big problems for Animators and Mocap guys. Even for the good ones out there, they can only do so much work being 1 person no matter how skilled they are. This new AI that Wonder Dynamics has will be able to do the work of an animator, lightning guy, rigger and character artist (in terms of blockout that is) all in one. Assuming the AI has competency at doing common animatable poses/actions (i.e. walking, jumping, talking), WD pretty much eliminates the need of 3-4 3D guys for any company or product that just needs basic 3D stuff you see in commercials or very simple games.
Lots of games come out dead on arrival, or are cancelled. You can be the best artist in the world, but if you're working with a studio with programmers and designers that don't know how to make an appealing game, or if you have a bad art lead/director, you're basically SoL. These people will end up using AI, and not only will their game suck, their art will too.
It all smells super fishy to me. I think right away of Hollywood accounting but also how difficult it is to discern just how much of any input data is actually used by an output.
Plus, the datasets will continually be diluted, both with other artists, but also with machine-created data.
As for AI, I mean we're talking about collection of shared data that somehow ended up with a stable diffusion, it is only as good as that shared data collection. As an indie game developer I have seen few things that AI can not do. Like, write a complex code and distribute it amongst multiple files, or create an image that exactly fits your imagination. These two factors will never be replaced but results from AI can always be used as references or some guide that may help you resolve your coders block or artist block, which had happened to me in the past.
In short, good tool, but extremely unreliable, just like the rest of the technologies.