|
I have some Thoughts. Chapter 1: Whose Fault is it Anyway? Some core philosophical concepts people should keep in mind: Technological Determinism: The idea that a technology that comes into being will inevitably proliferate across society; it is impossible to effectively regulate, it will spread, and change all lives it touches. Karl Marx one of the most prominent technological determinists further expands on this to suppose that the inevitable proliferation of technology changes society which results in the development and proliferation of new technologies and so on, "the Sorcerer's Broom" etc. See the possibly apocryphal account of the Catholic Church trying to ban crossbows. Karl Marx writes: quote:Meantime the markets kept ever growing, the demand ever rising. Even manufacturer no longer sufficed. Thereupon, steam and machinery revolutionised industrial production. The place of manufacture was taken by the giant, Modern Industry; the place of the industrial middle class by industrial millionaires, the leaders of the whole industrial armies, the modern bourgeois. It's easy to see how technological progress and its spread maps onto the above. Social Constructivism: The idea that technology exists at the behest of humans, is impacted by humans, for humans, is affected by the pre-existing social context, and can be regulated by humans for humans. The historical example here might be Ming China or the Shogunate of Japan. Some of the debate will shaped by whether you fall into one of the two camps or somewhere in the middle between them. Personally I lean towards Technological Determinism; too many corporations and Universities and research orgs, including nation-state actors viewing AI as a new arms race have invested a lot of time and money across the globe into AI. There are too many day to day benefits and applications that add to GDP to just not join the race. A similar situation might be stem cell research, the US attempting to ban it merely let other nations get a head start in developing the technology. The Stakes: Many people are rightfully scarred and concerned about the technology; the massive and unprecedented amount of what is at least arguably a form of theft by data scrapers to form the datasets to train the AI, is concerning. But I kinda feel like some of the online debate and concerns by artists are kinda missing the forest for the trees. Because assuming we do legally settle the issue in favor of artists regarding data scraping that doesn't prevent the AI from eventually getting good enough through a long process of ethically sourced artworks in displacing artists. There's also the issue where honestly? I'm not sure what the difference is between an artist potentially losing their job and horse drawn carriage drivers 100 years ago losing their jobs, or the craftsman, as Karl Marx wrote 200 years ago: quote:The bourgeoisie has stripped of its halo every occupation hitherto honoured and looked up to with reverent awe. It has converted the physician, the lawyer, the priest, the poet, the man of science, into its paid wage labourers. quote:The lower strata of the middle class — the small tradespeople, shopkeepers, and retired tradesmen generally, the handicraftsmen and peasants — all these sink gradually into the proletariat, partly because their diminutive capital does not suffice for the scale on which Modern Industry is carried on, and is swamped in the competition with the large capitalists, partly because their specialised skill is rendered worthless by new methods of production. Thus the proletariat is recruited from all classes of the population. This has been a process that has been ongoing for hundreds of years since the advent of capitalism and the rise of the bourgeoisie. No regulation or law is going to stop capitalism as a socio-economic force from destroying previously unassailable modes of production. The real solution is probably the adaption of some form of socialist economy; whether the basic needs and necessities are met and large corporations wouldn't exist to devalue the work of craftsmen out of a desire for endless infinite never ending profits. The technology can be used for its good, and not merely for what it devalues. The problem is always capitalism. Chapter 2: How much of a problem is it currently? But personally based on what I've observed as someone who commissions a lot of art, I think artists in general are maybe a tad premature in their doomsday prognostications. I don't doubt the technology might get good enough to be hard to distinguish from a human's hand, but I think this doesn't capture the full nuance of what's going on. For one thing, insofar as AI Art is still flawed, many artists I've seen resorted themselves to a new niche, as "AI Art Retouchers", taking jobs to fix the AI art and making it "gooder". It's evidence to the counter argument (by liberals typically) that new technology doesn't "destroy" jobs but creates new ones!  Putting aside that its unclear what the ratio of old job destroyed vs new job created is; this is a something to consider. Also most of the AI Art I've seen falls into a broad but also paradoxically specific category of AI Art. People Who Don't Know Any Better Without Taste Just Wanting Their Pretty Gaudy Art(tm). Seriously most people who seem satisfied with AI Art are just extensions of "Look at this weird funny result the AI made, haha isn't that quaint?" to "Look it's my special snowflake Sonic OC! " Generally a category of people who probably didn't have much of a budget and probably don't account for a significant share of art commissions (in total value). For people who actually have a more specific desire, whether it be pose, style, props, costume, and so on; the short comings of AI get really tight and I'm not sure this is something that can ever really get resolved. As someone with their own indie game project, the few times I've tried using the AI the results were so atrociously unuseable, for as reference material *for* a artist I intend to pay. I think large corporations might even be in the same boat, you're not making the next Halo with AI art and expecting it to be critically acclaimed; many will of course try, and fall flat on their faces but I think there IS a limit to how much of a soulless cash grab a game can be before customers and audiences have enough. Where things get tricky is the Indie scene, I've seen AI get a lot of usage already; whether it be for code generation (no ethical issues here honestly, programming is just copying and pasting with extra steps) or placeholder assets. I think the use of AI to generate placeholder assets or aid in kitbashing is clearly something we can see helps some field a lot, but is it at the expense of artists? Can it be done to *not* be at their expense? Can a compromise be met? For example, using AI Art to generate placeholder assets which once you've completed the functionality of the game go and pay artists to generate the for-release assets, stripping out any assets you can't afford to replace? As someone who commissions art, my process currently is this: Write a Description of what I want, then I go onto Pinterest and Doom Scroll for between 4 and 72 Hours downloading reference images (thonk emoji, I wonder what this step reminds me of?), which I then sort and pick through and then I kitbash the result in a document or an image file which I email to the artist, sometimes with addendums. This is a process that took me many years to polish and is basically its own skill because the artist doesn't want to have to spend a lot of time revising the output right? And I don't want to have to spend more money than what the listed price was. And as I wrote in the TTRPG thread I think like Piracy, this feels broadly similar to a Convenience issue. A lot of people out there who go out and seek artists for commissions have no idea what they're doing, get frustrated, and this forms the core of the current market for most people who are currently into AI. You need a combination of basic social skills, an eye to distinguish style and skill, and a basic understanding of the limits of your budget and an idea of how to communicate your vision clearly and concisely to the artist. Most people who go out there getting their commissions are like stupid teenagers with their birthday money or allowance who as I said, no idea what the gently caress they're doing, get easily scarred off or get their heads bitten off because they lack tact or something or talked to an artist who was having a bad idea, and so on. I'm not going to say its the fault of artists, but a lot of artists out there aren't helping themselves, like by not listing their prices! So gasp someone now has to ask you for your price, that's asking for trouble and is too much for most people. I remember one artist who said in an appeal against using AI art, "Look if budget is an issue, artists will be willing to negotiate!" What the heck, no, that's not even a majority of all artists and many will get very understandably upset if they're being asked to lower their prices. And there's huge debate even among artists about what prices people should set their art at. There's no single set of expectations that will apply to every artist, I discover everyday where I go out and commission new artists just new and all sorts of small or big ways artists differ from each other, like even in things like basic terminology. Some artists expect references, one artist I met fell onto their knees and cried that I gave references up front and were shocked, every artist is different and that creates a kind of problem for a lot of people who understandably dislike being asked to run through a minefield in order to spend money. So it is tempting for me to consider AI for generating references, instead of spending *hours* finding pre-existing art with the exact pose I want (albeit this is something programs (hrmm) exists already for), the costumes I want, the props I want, and so on with just a few clicks and maybe at most an hour of my time, to then still pay an artist anyways; which lets me better strategically spend my limited budget; that's a very enticing use case; I haven't personally experienced any AI that can do this that isn't a huge hassle that's more effort than its worth, but I can hypothetically see the use case where ideally the dataset that trained the AI was ethically sourced but I also don't really see a big difference between what I described and scrolling through pinterest? Chapter 3: TBD, Solutions aside from the complete overthrow of capitalism?
|
#
¿
Mar 25, 2023 16:24
|
|
|
|
| # ¿ May 10, 2024 09:02 |
|
|
I think the main use case for a "legal" AI is scanning through documents from Discovery, which as I understand it is the most expensive and tedious part of lawyering.gurragadon posted:Thanks for the effort post on AI art, interesting to read. You said that AI art is atrociously unusable, and a lot of artists are trying to find work touching up AI art. What is it about AI art that you see that makes you recognize it so readily? Or is it the struggle to communicate effectively what you want to the AI program? You mention that people use it for "weird" stuff, but is there a difference or AI'ness to a picture of a sunflower or something else generic? There definitely will be a few artists who make a living retouching AI art but its whole point will lead to less people being able to support themselves as artists. To be clear "weird" is more referring to early AI use, when most artists probably dismissed AI as a threat (except for a few exceptions). Anyways, maybe there's other better programs than the ones I used, but trying to specifically spit out anime or manga style designs just resulted in unusable outputs due to like, massive issues like mismatched eyes and the AI stopping and starting different styles like a picasso painting. From what I do know from people who do get better results it seems like it involves turning a lot of knobs, fiddling with parameters, uploading reference images to use as a base and so on which I feel like defeats the purpose; its me needing to learn to use a different sort of tool and spending probably a nearly equivalent amount of time. Most importantly though I feel like maybe if I know very specifically what I want I'll get it, and no more. The advantage of being stuck in Pinterest Hell of having 150 tabs open as a result of clicking through and opening "promising" and interesting results is I end up with a LOT of potential sources of inspiration that I subsequently download and categorize (heh, just like them robots) for future use. Ultimately there is a certain enjoyment to this process as well as frustration which is I think the crux of the thing; when AI is enhancing workflows great, if it is helping artists with doing tasks they'd rather not do, great. But when it infringes on parts of the process and the craft artists enjoy doing? No thanks. And the unfortunate thing is a lot of the "AI Bros" on twitter just seem to be very out of touch about that part of what makes learning a creative pursuit so satisfying. I dunno how to best formulate this thought but there's a certain observation I've made and I'm not sure how to express it that I feel like captures the essence of AI popularity at least in the anime community. There's a certain "style" of what I tend to think of as Overpolished? Anime art where the lighting, shading and so on seems like its pushing the art to its limits but is otherwise uninspiring in terms of pose, anatomy, appeal and so on. There's one artist off the top of my head where I see their art and I think "That's very technically impressive but still kinda... Soulless?" And this spawned a lot of a sort of imitation style of similarly overly refined works that all kinda blend together and look the same... JUST LIKE AI and results in the unfortunate trend of some artists being accused of AI art because they also kinda adapted that style, a style that I think is largely born out of the kinds of anime/manga artist who self-teaches themselves anime from other anime artists and never got formal art education for things like anatomy/composition/etc; just brute forced themselves to a certain level of visual appeal and quality from memorizing the 100s of lighting/shading tutorials and using anatomy posing tools as reference. For years I felt I could tell the difference between someone who was self taught from anime, self taught but took the time to learn the fundamentals, and a formal art training prior to anime art and it speaks volumes in just how much of a difference it makes in making appealing characters. That I can't really tell this style of 'self taught from anime' anime art from the AI anime art I think anecdotally kinda says something; on a very broad level I mean. I don't wish to generalize of course and I'm not sure what it says; but the way they seem to pair and complement each other is certainly very interesting to me. But yeah regarding retouching I think the main point is I think the people willing to pay for art will continue to do so, and the people who whine in artists dms about how unaffordable they are never did and never will. Like... I primarily see AI art as something that has its biggest audience in the indie adjacent space of "I wanna make a card game like magic the gathering" or "D&D 3rd party content book" with "But I can't afford an artist.  " where hypothetically the AI is displacing artists but I don't think its displacing very many. The people willing and have the means of paying hundreds of dollars for MTG style/quality art that the AI kinda excels at producing are probably still willing to pay that amount; its the people with sub-100$ budgets who are going to value the ability to self-publish their dream supplement/card game and push on ahead who never had the budget to begin with or only had the budget for art that's basically below minimum wage broken down per hour.quote:Alot of the confusion you describe with artists seems to be that they have their own way of communicating and it's not laid out very well to people who are unfamiliar with commissioning art. Like you said, I wouldn't have been prepared with reference photos of what I wanted unless I spoke with the artist beforehand, and they told me they needed that. I would have expected to tell someone who is claiming to paint or draw what I wanted, and they could do an ok job of it within their vision if I just told them. That seems like more of a problem with artists being unable to communicate with people outside of their field well, which is not rare by any means, but it's pretty essential for the artist to be able to. So putting aside language barriers which is definitely also something I sometimes would come across, its a little crazy to me the idea of trying to ask an artist to draw something with just a written description and nothing else. The superposition of all the possibilities just seems so endless and infinite it would be any wonder the final result remotely resembled what you wanted. But there are some where this is the expectation, I've heard of artists who actually insist on a MAXIMUM of TWO reference photos! Two! That's crazy talk, I provide on average 15 for every aspect of the design to insure the least number of potential revisions while still leaving some wriggle room for the artist to present their own style and vision and input to the process. So this is a part of the problem, communication goes two ways, clients who don't know what or how to communicate what they want, and are lacking in the experiences to know how to or for how long to be sorting through artists and portfolios to find one that suits them; and artists who need to better present themselves as a business and convey clearly their process, their prices, their portfolio, and expectations of you as a client. quote:It also leans into the question of the artist not recognizing the vision of the person who is commissioning the art. Art is subjective and while there are things that are considered "good" and "bad," maybe I just want the artist to draw me something that they would consider "bad." Do they want to do that? Probably not because their name is going to be on the piece. But if they are making art for business then they have to abandon their vision as an artist. It's why I think introducing money into anything cheapens it by some degree. AI art doesn't care about what's "good" and "bad," it "cares" about spitting out as many images as it can. Usually artists indicate a list of "Do's" and "Don't" in terms of content they draw. It is I think pretty rare for an artist to be presented with a request they morally or stylistically object to; I think some artists I know told me about how a client would pick and insist on absolutely garishly wrong colours that just hurt to look at; I can only kinda vaguely remember an artist suggesting a change to the design but usually I actually do defer to the artist; it's why I'm paying them, to use their skills and knowledge to tell me what parts of the design just won't work, where the physicality is all wrong etc. So generally most of the time I expect an artist to refuse a commission they aren't comfortable with, I've certainly seen artists react negatively when I asked them like, "Hey do you do anime art?" when they're a semi-realist/realist painter, so I think typically they're very willing and upfront to tell inquiring clients to gtfo if what they're asking for isn't on offer.  The fact is there's so many artists out there, that it isn't difficult at all to find one who does what you want... If you put in the time and vaguely know where to look. Someone commissioning their OC for the first time probably doesn't know where to look and doesn't have an excel sheets of artists they find promising. There's actually a reason why I track artists which is to in particular remember which artists I enjoyed working with who seemed genuinely interested in the Lore(tm) of my characters/world/game idea/etc. Not just because if I ever got the budget to hire them they're on the short list but also because I'd like to revisit and commission them again because they like my ideas and are clearly firing on all creative cylinders working on it. Which again I think wraps back around to what's motivating the AI Art people, they mainly experience very negative interactions with artists, they're like what I'mma call Artcels; they have a lot of trouble connecting and vibing well with artists and so basically want to take their OC ball and go home and the AI fills that void which is basically in a way, a lack of social interaction and connection. quote:I guess the difference in AI and scrolling through Pinterest is you can generate images faster with an AI program, and if you change your mind halfway through it's a lot easier to just start typing in new terms. AI art generated through a program is also an original composition based on other images, not just the original piece done by someone on Pinterest. So maybe I would feel a little less bad about copyright, but that dosen't seem to hold very well and I don't really feel bad about copyright law anyway. Pinterest in many ways is pretty similar, I can type in new terms to try to massage the results to be closer to what I want, like "female mage" gets some kind of result and then I gotta click around to find a result that looks in the ballpark and then scrolldown from there. Pinterest has I think probably a similar but more archaic version of the sort of "referencing" and linking together of ideas I think, like the way you click on a mage with a red dress and suddenly its all Chandra from Magic the Gathering. But yeah to talk more about the idea of "Experience", originally my budget would be pretty low, but gradually I begun to realize I was really spending a lot more than that in actuality. Because often I wouldn't be satisfied with the design, one thing would be right, other aspects wouldn't be. So I'd find a different artist and try again. How many times does this take before I was satisfied with the result? I think there's one character I legitimately have like ten versions of before I got to like final design I was happy with. The lesson I learned eventually was I was ultimately budgeting too low; and got very varied results, if I wanted better results I needed to pay before, and gradually this increased. I think I now spend closer to four times my original spending limit per character and basically get perfect results. Sometimes I see someone on twitter who is like, "Full character designs for [absurdly low price]" and I see "Huh they're art is actually really good especially for that price, I'll give them a chance." with the knowledge that it's technically a risk and sometimes I get surprised at how good it is. The point here is its another facet of the commissioning process that can be very fraught, where people also don't know how much this service should be priced at, so I can easily imagine some people who got disappointing results early on and see the AI as something that gives them "good enough" results for that hit of dopamine to not bother learning how much they are actually willing to pay to get the characters brought to life "perfectly". This isn't to say of course that any of this is particularly means its justified to just blindly use Midjourney or whatever because someone had a bad experience one time talking to an artist; but mainly to posit I think there's maybe a complex web of circumstances regarding the artist/art commission industry that contributed to the popularity AI has which can frame it potentially as being a service driven problem. Raenir Salazar fucked around with this message at 05:24 on Mar 26, 2023 |
|
#
¿
Mar 26, 2023 05:19
|
|
|
One thing I wonder about the new discoveries chemistry chat is how it compares to the people who did discover new treatments because they got access to pay walled information via Aaron Swartz work ( ) and I wonder to what extent the ai could speed up processing of such data by outside interests. But yeah, back to ai art chat for a moment, for me definitely like, the entire point of commissioning art is the interaction. AI doesn't fulfil the void in me for human interaction, and interacting with artists who seem genuinely interested in my ideas gives me quite the dopamine hit. I can't imagine and am utterly baffled by people who are satisfied with using the AI completely because it's just so incompatible with the ideas at play for me. As a pragmatic practical cost benefit analysis matter sure I am intrigued by ways ai can save me time, or let me find references I'm having trouble nailing down, but they could never replace the role the artist actually has, to be my captive audience.  I actually get a little frustrated when the artist is *too* professional and standoffish because then *I* get no feedback as to whether my design is cool or interesting and I actually think I have a minor talent in coming up with character designs from doing it so often, and interacting with an interested artist let's me further hone this particular skill, I need that feedback.
|
|
#
¿
Mar 27, 2023 17:13
|
|
|
Rogue AI Goddess posted:Make it obsolete with Capitalism 2.0. There's no reason why an AI can't be trained to create and run a business. It'd outperform musk probably.
|
|
#
¿
Mar 28, 2023 14:53
|
|
|
SaTaMaS posted:The thing keeping GPT from becoming AGI has nothing to do with consciousness and everything to do with embodiment. An embodied AGI system would have the ability to perceive and manipulate the world, learn from its environment, adapt to changes, and develop a robust understanding of the world that is not possible in disembodied systems. How do we know we're not also perceiving the world via a disembodied system though? What is embodiment other than a set of inputs? If we provided a sufficiently realistic and informative simulated environment how does this differ from the Real(tm) environment for the purposes of learning?
|
|
#
¿
Apr 8, 2023 20:20
|
|
|
I was doing some coding today for an Unreal Engine project and definitely came across a situation where it spat very convincing nonsense at me. But ultimately when I tried to use the code it gave me the functions basically didnt exist or it provided a solution that just didn't work or seem to actually comprehend the problem. And I think in this case it was because there might not actually be a solution for what I wanted.
|
|
#
¿
Apr 8, 2023 23:26
|
|
|
Doesn't the "is it copywriter infringement?" question also have a sort of exception built in regarding what the states interest is? I wonder where even if a judge decides "well it's technically infringement but ruling that way would overly negatively effect the United States in regards to the development of new and novel technologies..."?
|
|
#
¿
Apr 10, 2023 15:26
|
|
|
From a game design standpoint it'll be really interesting to see if anyone tries to revive the "type in instructions" subgenre of adventure games. Using AI to interpret the commands more broadly. At a minimum if it's supports speech to text, might lend for more accessible games?
|
|
#
¿
Apr 11, 2023 20:51
|
|
|
|
| # ¿ May 10, 2024 09:02 |
|
|
IShallRiseAgain posted:I'm working on that right now. https://www.youtube.com/watch?v=uohn5o0Cgpw Heck yeah!
|
|
#
¿
Apr 12, 2023 01:05
|
|