This entire article misses critical developments in the AI art space.
Controlnet was just released last week. It allows for precise fine-tuning of the image with skeletal positions, depth maps, outline sketches, etc.
It has been exceptionally received in the AI art community, because pure-text-prompting precisely runs into the issue described in the article. You can only convey so much information with text prompts.
That being said, artists in general still don't want to touch controlnet. Because despite controlnet solving their complaints about how the AI isn't controllable precisely, it doesn't solve the real problem, that drawing skills are massive devalued.
Artists will still exist, but most likely as hybrid 3d-modellers, AI modelers (Not full programmers, but able to fine-tune models with online guides and setups, can read basic python), and storytellers (like manga artists). It'll be a higher-pay, higher-prestige, higher-skill-requirement job than before.
And all those artists who devoted their lives to draw better, find this to be an incredibly brutal adjustment.
PS: Despite how people made fun of 'proompters', or predicted that prompting would be automated away. The skill ceiling to good AI art has radically increased. There's now 10 different fine-tuned models you need to learn the basics of, each with different strengths. There's thousands of LORAs to insert into prompts to precisely reproduce a subject that the base model has no information on. There's 20 parameters to tune, each with different effects. There's VAEs which affect coloring and fine-details. Now there's controlnet, and you'd better learn blender to rig basic skeletons to feed it.
This likely suggests the future of an AI-augmented economy: People become AI wranglers. Wordpress was supposed to make blogging easier, instead it spawned an industry of WP wranglers.
This is the critical insight. Human artists having nothing to worry about. Human drawers do.
The decoupling of concept from execution is old news in most forms of art. Writing music and playing an instrument are two very different skills, and indeed any session musician will tell you that playing instruments well is not as valued as writing amazing music.
In fact we went through a similar thing in pop music in the 70’s and 80’s, with the brief moral panic over electronic music and people who “just push a button and don’t even know know to play their instrument.” Turns out nobody cares, as long as the music is good.
AI art decouples concept from execution for visual art. That’s all.
You're not taking into consideration how the actual artists actually feel. And, from the first hand - the feeling is horrible. Most artists love to draw, the process itself is incredibly satisfying. Just like a musician loves his instrument.
This is destroying them, stealing the joy from the thing they devoted decades of life. My wife is extremely depressed by this, to the point I think she'll need serious therapy. She still has most of her job, but yeah recent developments like ControlNet? Well, shit.
"This is the critical insight. Human artists having nothing to worry about. Human drawers do."
The reason why amazing music is valued so much is because there's so much music that you have to be amazing to be noticed. Art has had that problem for a long time. Someone spends 100hrs of talent on a masterpiece and we say "meh, seen a thousand of that quality"
Learning theory is easier than applying it so becoming an artist is easier and now artists have more competition. Not advocating against AI art but it's obviously going to have a negative effect.
Even human drawers might do all right, if they're selling physical drawings. People will still put hand-painted canvasses on their walls, just like people still enjoy an acoustic guitar concert, even if the guitarist isn't world-class. The authentic connection with a human being is where the art survives. If you're not an amazing artist, you can still be an authentic one.
But if you're just churning out commercial illustrations, then sure, an AI can do that now, or soon.
> Turns out nobody cares, as long as the music is good.
Actually, all those people who lost their professions they spent their lifetime working on, they care, rather a lot.
> indeed any session musician will tell you that playing instruments well is not as valued as writing amazing music.
Yes, in 2023 they say this, but I assure you when "session musician" was a common profession, that no one said this.
When I first started in music, if you wanted your piece of music, you had to pay instrumentalists to play it, and this was a valuable professional skill.
> Turns out nobody cares, as long as the music is good.
And the music is not good either.
When did everyone become so sociopathically detached from the welfare of others? It's just horrifying.
I'm not an artist, but my fiance is an animator. Generally, something similar exists in stock assets (photos, assets like vector files, other such things). We had a discussion on whether the existence of stock assets means drawing is no longer necessary to be an artist, and her answer was it still is something you should be able to do because knowing how to draw is the best way to learn the fundamentals of art in general, knowing how parts of a body work in a drawing or animation, and generally something is "good" vs. "bad," etc. However, as someone who isn't a fine artist, she doesn't have time to generate everything frame by frame, so she doesn't do it for every project she works on, but that knowledge is invaluable and one you can only obtain from knowing how to draw.
I think the best analogy I understand it is as is it is like assembly language or low level programming in general. I certainly do not have the time to program every piece of code I write in asm but having done projects in asm is invaluable as a coder[0], given how much it informs my mental model of how the code I write actually works. That understanding is beyond valuable and is something that puts you a rung above everyone else who just copies things from SO without knowing what they do. I think AI for devs, to the extent that it will evolve, will still be as such. People who primarily find it amazing today I find either 1) use it as a productivity boost for things that need a lot of boiler-plate[1], or 2) are SO-copy-pasters who are just amazed they have to think even less. A LOT of the "AI artist" community are the artist equivalent of the 2nd, honestly, and it's easy to detect AI art because it's generated by people who are not really artists, and in similar fashion either don't know the fundamentals or who only create "good" work by nearly directly copying other art pieces.
Also, to critique your analogy, what you guys are saying is along the lines of "given DAWs, why learn to play an instrument at all?" Except a plain look at any good composer will show they know how to play at least one instrument, even if they don't play all their music on that instrument, there is no doubt knowing how to play clearly makes you a better composer. I mean, can you even imagine a composer who cannot even play piano, or guitar? Sure a composer need not be a virtuoso concert pianist, but they should at the very least be able to play the chords of the very song they've composed. Nothing thus far teaches the human mind deep understanding of something more than doing that thing does. It is so clearly obvious in music and programming and the only reason people keep saying you can be an artist having never learned to draw is because such people simply do not understand art or composition at all.
[0] I'm not really a software developer but a computational scientist, and so I won't say I'm a developer, but still understanding of data structures, algorithms, discrete math in general, and yes asm is invaluable for me.
[1] As someone outside of the dev space, I don't know why you people don't just make better interfaces instead of juggling 9 yamls and 3 environments for every project, or just roll your own interface and reduce the boiler-plate yourself.
> Because despite controlnet solving their complaints about how the AI isn't controllable precisely, it doesn't solve the real problem, that drawing skills are massive devalued
At this point if I was an artist I'd probably be looking for a way to leverage my existing skills into AI art, because it seems like drawing skills will go the way of the horse and carriage.
I feel bad because this has happened so quickly. Previously it seems like there was a longer transition period.
This kind of transition has got me thinking the same about coding though. If AI programming does indeed take over, my programming skills will likely be massively devalued.
If you believe this is going to happen, how should you prepare for it? Start learning prompting now? Get involved with building these AIs? Etc.
I've tried to convince artists to transition to the AI-augmented future, since October.
There's almost no successes. I've seen artist communities on multiple websites, in multiple languages. At first they tried to laugh at the AI. Now the AI has improved so radically so fast, they universally prefer to stick their head in the sand, and ban discussion of AI altogether, its all denial and rage.
On the more optimistic side. I see an extraordinary explosion in artistic innovation, just look at websites like CivitAI. The massive community all training subcomponents of the models for each other to share. The models rapidly improving every month just through fine-tuning and theoretical innovation, without stabilityAI's involvement (They are distracted by lawsuits now).
There are many 3d-artists intensely experimenting with AI art, to say make AI-anime, which has illustration qualities on every frame (A previous impossibility due to the costs involved).
It seems with AI, it'll really cleave communities in two. The ones who eagerly embrace it, seem to enjoy it extraordinarily, and achieve quite a lot of popularity and success. But the rest just want to pretend it doesn't exist, waiting till employers realize that they are no longer needed.
Regarding programming, it doesn't appear that AI programming can replace humans. Programming is very similar to novel writing in terms of complexity for AIs. And AIs are still extremely terrible at long-form storytelling. The lesson is to aggressively use AI tools as much as possible, to understand the long-term weaknesses of AIs, and deliver your values in those areas as a human.
> seems like drawing skills will go the way of the horse and carriage.
The exact same thing happened to technical drafters when CAD destroyed the entire (sub-)industry.
And to typesetting, for that matter[1].
Which is a pity in the human sense and the sense that both were forms of artistry and produced things of great beauty, both the product and the machines used to enable them, but they're simply not economical in the face of Solidworks and digital composition. And yet the replacement technologies have also enabled a lot more creativity and further advances. Objects with complex geometries are now possible to specify and manufacture, when they previously could not even be accurately drawn.
1: https://en.wikipedia.org/wiki/Wapping_dispute, where not only could the unions not simply oppose the rising sea-level of technology, but their failure seriously damaged union credibility in general.
> At this point if I was an artist I'd probably be looking for a way to leverage my existing skills into AI art, because it seems like drawing skills will go the way of the horse and carriage
Drawing skills were made redundant for producing most types of high quality image before the motor car but people still pay for hand drawn items
Not sure AI lowering the skill barrier for and speeding up the generation of digital art is really going to change that
> my programming skills will likely be massively devalued
I see this in the context of the drawing vs. art analogy. Yes, your typing and maybe syntax skills will be devalued, but your higher level, creative programming skills are probably safe for a good long time yet.
>If you believe this is going to happen, how should you prepare for it? Start learning prompting now? Get involved with building these AIs? Etc.
If you still want to work as someone who produces code - except your personal code factory has changed from brain and fingers to AI - then yeah, you should probably do both of those things.
Even if you spend 1-2 years and this AI hype doesn't work out, you still have those skills to go back to.
Frankly, this mainstream adoption by Google and Microsoft is, uh, not going great. So you can afford to observe for now, but AI advancements have been made very rapidly, so it's not wise to completely ignore it either.
> If you believe this is going to happen, how should you prepare for it? Start learning prompting now? Get involved with building these AIs? Etc.
Short of advocating for artists' unions...
Law is still going to govern anything AI creates, so anyone who develops a skillset in both art and law would be well-positioned to gatekeep as a copyright troll. This carves out a career for yourself and enacts revenge by making it a liability for employers to use AI to replace artists.
HR is the theocratic version of that, since law school isn't cheap. You can still gatekeep, but you'd be playing by more-arbitrary rules ("you can't use that AI because it incorporates images of Women Without Penises," etc.).
Security is pragmatic, but I worry about the long-term stability of it. Every time a breach is announced, there are no consequences, so why even pretend to need it?
I used to suggest pivoting to tangible works and experiences (architecture, sculpture, etc.), but those are easily displaced as well-- I would bet against it now.
Look at porn-- what started as a handful of performers in studios is now done by any college student with a webcam in their bedroom (next up to be replaced with AI-generated content).
For now, you can still add value as a sex worker by offering the GFE, but even that's about to be obsolete. You can get your fix of flirting from a chatbot (TTS or sexting) and 3D print a copy of anybody else's genitalia. Remote-controls, vibrators and fluid pumps add some life to it. And you don't even have to leave the house, which affords you privacy to pursue darker subjects without oversight and save you more time for repeat consumption.
Go with the flow. AI will need humans to be effective, humans will need AI to remain competitive. But everyone will have the same base models, just like we all have the same web search and electricity. AI won't be a competitive advantage, it will be a basic requirement.
Do you believe the number and complexity of software applications will decrease in the next 10 years because of AI, or that it will spawn whole new ecosystems of software and new types of jobs? I believe the second is more reasonable, we will have higher expectations from software in 2033 than in 2023. The easier AI makes it, the more difficult we make the tasks.
Human desires fill the available space like air, AI exponential is slower than our entitlement. So we still need to work.
I believe this is going to happen (and is happening right now, in front of us) not just for programmers but for any role that can be automated.
> How should you prepare for it?
Get equity. I mean ownership stake. If you don't have a legal/financial claim on the output of the machine then you're about to be part of the worthless surplus from the POV of the system.
> Start learning prompting now?
No, the machines will do that well enough in a minute or two.
> Get involved with building these AIs?
No, the machines will do that well enough in a minute or two too.
- - - -
Artificial Intelligence destroys scarcity, which is the fundamental basis of our societies and economies.
Think about it: scarcity is the very problem that societies and economies evolved to solve in the first place.
Now science and capitalism have delivered technology and wealth. There is enough to go around if we just worked out the logistics, and computers can do that for us in a matter of moments. In other words, the "World Game" is not hard! https://en.wikipedia.org/wiki/World_Game we just have to get over our hangups.
Now here's where it gets really interesting: ChatGPT et. al. don't have glands, they don't have emotional trauma, no PTSD from being humans-on-Earth for generations, etc.
We can program them to be sane and perhaps even wise.
We can also attach empirical feedback devices to them, make them scientists...
So we have physical abundance and benevolent, sane, empirically-grounded AI advisors, how much longer will it take to sort things out? I think we could be looking at the start of a Golden Age?
Yeah this post seems like it will not stand the rest of time.
The fundamental conceptual shift is the emergent behaviors displayed by these models. And that is highly unforeseeable.
We’re not working with machines that do what they’re built to do. They’re doing more. And every day we discover something new they’re capable of doing.
Making any sort of statement about the limits of this technology is going to be shortsighted.
The opposite is also true. People overestimate what is possible and underestimate how much more effort and time is required to start replacing industries.
> Artists will still exist, but most likely as hybrid 3d-modellers, AI modelers (Not full programmers, but able to fine-tune models with online guides and setups, can read basic python)
Why should artists have to learn fucking Python!? There are so few jobs open to people who don't want to "learn to code", this was one of the last, and that's going too.
> This likely suggests the future of an AI-augmented economy: People become AI wranglers.
So we don't get to create ourselves anymore. No drawing, no making music, no programming even. We just give prompts to AI programs. Which means we are worthless, because any bozo can do this.
I hate this timeline so much. It's like all the worst and most mediocre parts from every SF future, with an extra dose of stupidity and greed on top.
I wouldn't worry too much. We've pushed the biosphere past the point of repair. This "utopia" they dream of won't have a chance to come to fruition before we all go down in flames.
And given this dichotomy as choice, I would choose the flames too.
At the end of the day, we are not going to see AI art in the Guggenheim, or any of those other museums where the finest art in the world resides. AI art is and will always be a novelty. Real art comes from the human spirit. It is not about technical execution, and not just about painting a pretty picture. In actual art we want to see humanity, something that AI will never, ever truly know about.
Seems like a lot of people are missing this point, and it seems like a lot of people have a chip on their shoulder about being unable to create art, and believing they can express the deep recesses of their feelings and psyche and everything else that makes us human by proxy through AI, and thus completely missing the point of creating art in the first place.
These are great platitudes, and as a fellow human I like the idea that only we can create "real" art, but I don't know how to justify that idea. I have no chip on my shoulder about art, it's not part of my identity at all. Computer programming is, though, and I fully expect it to affect my industry in the next decade.
The current fine art market is extremely unmeritorious so I agree that AI art will never make it to the Guggenheim. That isn't a good metric since the human curators will never allow it to happen. The metric will be the industry. Will ai start composing scores for movies? Promotional posters? Art that's sold outside of expensive curated galleries (etsy, ect)?
It's possible that humans will keep their stranglehold on these and I hope that's the case... But I wouldn't bet on it.
I think even without AI art we are well past the point of art in famous galleries making sense. Isn't most of this just thinly veiled money laundering anyway?
I imagine we WILL see AI art in a famous gallery for exactly this reason. Except it will be lauded for the imagination of the creator and the tweaks they made, not their drawing ability.
> This likely suggests the future of an AI augmented economy: People become AI wranglers. Wordpress was supposed to make blogging easier, instead it spawned an industry of WP wranglers.
Good counter for "AI will steal our jobs". When competition starts to use AI+human, you got to level up. And since everyone has the same base AIs, the differentiating factor is still human.
Artist here: Article is an idealistic take not based in reality- the Jeff Koons of the world will be safe with their 30 million balloon art sales.
Everyone else is going to be rekt.
All of the work my artist friends do is for businesses- businesses trying to make the biggest profit. Ai art generators are faster/cheaper/more flexible. The artists are done and now it's all about the prompt engineers that don't need to be artists to excel.
Market will decide what to pay "artists" but seems like it will be a race to the bottom. Artist types I know are preparing accordingly.
But you should realize that any artist learning prompt engineering should be a better AI-artist than a non-artist learning prompt engineering. After all, art is not jsut about technique but also creativity, composition, theme etc., right? And an artist can furthermore touch up the AI generated product manually if necessary.
Yep. I spent a good hour trying to generate some stuff with DALLE. I then asked a friend that works in media and he was able to give me ideas for a prompt that worked almost immediately.
AI art is fun and _sometimes_ it can replace some things but at the end of the day, I don't really want to mess around prompting over and over. If I pay an artist $100 for something, I couldn't care less how they make it. I feel like AI can be another tool. I'll even make a bold prediction. In 10 years, we will still have "Graphic designers" and "digital artists" the ceiling will just be raised similar to how it was when photoshop became more commonplace.
What leads you to believe those problems won't be solved in the next few years? What stops me from having ChatGPT provide feedback on my prompts to make them more artistic? Maybe I just pay some artist to touch my stuff up for me if they're desperate for work or maybe the output gets good enough that I don't.
Out of interest, how are they preparing accordingly? It's not remotely my own field but I'm starting to wonder how I should be guiding my children who have tallent for art, in light of all this.
Draw comics, or really manga in particular.
The only real challenge in AI art, that won't be solved any time soon, is storytelling: chaining multiple panels together for coherent stories.
Even ChatGPT cannot tell a coherent story longer than 2 paragraphs. Being able to write a long story well, requires understanding world modelling, human motivations , etc, AGI tier abilities.
Also, expose them to AI-art. If they lose interest in art after seeing AI art, that means they were never meant to be artists, they merely like to draw, not to make art. And drawing alone is not really economically useful anymore.
AI is shifting the balance of power from technical ability and quality of tools to creativity/vision (and ability to market). The only "art" job left is going to be something akin to an art director.
Expose your children to a wide variety of art, books, music, food, travel and so forth. Teach them why (aesthetically) good things are good and bad things are bad. Also, encourage them to create stuff for an audience so they learn how to present things, gauge people's tastes and become comfortable with failure young.
The harsh reality as far as I can tell is that the prospect of making a living as an artist, which was already slim if we are being honest, has now shrunk to near zero.
They should be learning about the AI field in order to create their art. That might be the only skill worth money when they are older.
Focusing on learning how to leverage these new tools to produce their own works faster/better/cheaper- focusing on their original ideas/designs/models- coming up with strategies to protect their source art from being used in other people stable diffusion models- learning about/utilizing nft tech to find new pathways to monetize their art-
In the end though its going to be all about original ideas imo- so creatives that don't have original IP really need to get on that and develop it- the new tools will allow anyone to create anything in realtime- people will be frozen by the prompt if they don't have original ideas/concepts/characters/worlds etc-
The temptation for those without original ideas will be to leverage chatGpt etc for "ideas" but humans are not required on that path-
So for your children- ignore the tools/technicalities and focus on ideas- original ideas-
And older artists said the same about computer art, things like :" this new artists do not know to mix paints or prepare a canvas".
Artists can adapt and use this new tools and do their job faster, you might have to sell a logo cheaper but you might be able to make them 20 times faster.
There will just be people who are employed to wield these tools, select, curate and combine these images into whoever commissioned the work.
I don't know who all these tools are really aimed at? As a software engineer, I'd still prefer to pay someone to mess with any type of "art" or design while I focus on other things.
I mean even if I had an AI program that I could use to code, to do SRE, to do images, to do accounting etc, I still think someone would need to be in charge or making these things happen, or else I'd just be busy prompting machines all day, which sounds mad fatiguing and boring.
I think the real question is what does "making these things happen" realistically pay, pennies on the hour to sit in front of a screen and maybe elevate that to the 1/10,000 person who has to come hit a few keys to get the station running again?
On the other hand, where was the backstop previously? Companies could have saved money on art by hiring unpaid college interns or teams from where the median wage is less than a few hundred dollars a month. Evidently there was some reason to favor hiring actual artists versus the cheapest possible person who can draw. Maybe that reason hasn’t died out yet.
My point is that unique digital art won't be replaced even by a hypothetical perfect AI image generator.
When it comes to the digital art workforce - it is likely to be strongly affected.
> Others might fall into a gap of “nice skills, but not yet that offer a business advantage”. Furthermore, the lower entry barrier to create any art is likely to result in the average quality going down - not unlike that plastic made manufacturing cheaper, but also less durable.
And you are right that, sadly, in many kinds of digital art, it will be a race to the bottom.
There's a mistaken assumption here. The author seems to believe that these generators operate on the basis of random chance, hence a 200x200 pixel image in RGB has a possibility space of 10^300000 images.
...But an overwhelming majority of those images are random noise, and are effectively homogeneous and interchangeable. They're perfectly identical.
Like human artists, the image generators are (very strongly) biased to create "structured" images, which are derivative of existing artistic works or natural representations in the form of photographs. In principle, there's no a priori qualitative difference between the AIs and humans in this respect -- no human art is created de novo, but is always a continuation of, or a reaction to, existing forms.
Further, there's no reason to believe that (a) the size of the possibility space matters in a quantitative sense, and (b) that AIs will never be as capable of qualitative "originality" -- in conception, composition, or subject matter representation -- as a human artist would be.
I put my caveats on indistinguishable images. And three is the challange - to make a better estimation of an effective number of distinguishable images.
> (b) that AIs will never be as capable of qualitative "originality" -- in conception, composition, or subject matter representation -- as a human artist would be.
Sure, AI already creates original things. The point is that questions like "create a drawing that presents friendship" can be approached from many, many different, and what is crucial - subjective, ways.
They are not worried about making all artists redundant, they are worried about making most artist redundant.
Instead of needing to pay for mediocre artist to get mediocre art that's good enough for a purpose, now the AI taught on the good artist (all subjective of course) can produce similarly mediocre art that's fit for purpose.
It's equivalent to replacing all of the simple CRUD web devs with AI that does good enough job
Interestingly, it might be easier to replace mid-skill artists than mid-skill developers (not to say both won't happen over time). The key difference is that when generating art, you can say "Make A in-the-style B.", get what you want, and move on.
Meanwhile, you have to live with the consequences of AI generated code.
Case in point: My wife was trying to build a system to generate POs per vendor from a series of workorders a couple weeks ago. She was using AirTable, which vastly simplifies the creation and use of relational databases. Given a lack of database experience, the structure that she made was misaligned with what she actually needed (and completely denormalized). Putting it into production would have caused major problems over time.
This made me realize that writing code is only one of the barriers to entry in software. Eliminating this barrier effectively gives everyone in the world access to a small team of incredibly junior developers who have no idea what they're doing.
Just like mismanaged DIY home projects create more work for professional contractors, I think AI might -oddly enough- create more positions for mid-tier and higher developers.
It most definitely is for simple reason - non interactivity. Image doesn't need to be told what clicking this button does or how that pop up menu needs to animate or million other interactions.
> Just like mismanaged DIY home projects create more work for professional contractors, I think AI might -oddly enough- create more positions for mid-tier and higher developers.
It might also create a lot of jobs for "AI crafters" - know enough about it to craft good prompts about what client wants and have enough graphic editing knowledge to tweak and mix AI input to create what is needed. It probably will also empower "one man shops" dealing in small customized websites for customers, replacing or augmenting what now stock images are used for.
Reading the comments it seems that everyone assumes that art is always and only presented as pixels on a screen. Whereas most of the art in my house is hand painted on canvas and wood, hand embroidered, hand knitted, hand woven, hand thrown and decorated pottery. One picture was hammered out of a sheet of pewter.
One of my favourite places is the Henie-Onstadt Art Centre sculpture park, another is Vigelands Anlegg in Oslo, no pixels in either place.
So why does everyone seem to think that art is only flat, lit, pixels on a flat surface?
I think the exact same thing, obviously because we're on hacker news where most people seem to be hoping they can spend their days prompting machines for art so they can replace "artists" and save some money.
It's actually funny when I think about it because most people wouldn't know what image to use and how to use it even when it was generated for them, there's even a skill in selecting art.
You could argue enough good photos have already been taken that all practically all photographers should have already been made redundant since the year 2000, we still have photographers.
I think if people are ignoring AI's limitations, it's more out of fear that programming is next than having something to gain. I doubt having the money/labour that's currently used creating art for something else would help me in a noticeable way:
-the money would just trickle up
-not that much money is spent on it anyway
-world becomes more depressing, nothing you look at had any effort put into it, no-one had to believe in an advert on any level, instead it's the output of a machine optimised to trick you (as one of a shrinking number of people with any agency) to spend and therefore make the machine stronger.
-artists/potential future artists decide to learn to code instead?
That reminds me of Ira Glass' quote about taste: artists know it when they see it.
Or why it's a mystery some folks prefer tabs vs spaces, or recoil from what they consider shoddy code.
Somewhere in the discipline of point, line, and perspective, in the composition of shapes and their organization, is the artifact of amalgamated neurons, to be observed by yet another consciousness.
I wonder if AI art will just help generate more gacha games. And one may wonder, of the limited time left on this planet, what really could we spend it on?
The library of every book contains no meaningful work in the search. The space of all generated art is oblivion.
Because that's what 99% of the people employed for their artistic skills are doing.
I doubt AI will have much effect on the tiny number of people who can make a living by producing art to hang up in your house. It's the people doing illustrations for magazines/websites/packaging/etc. who are fucked in the medium/long term.
People will 100% start hanging AI art on their walls. The killer feature is being to customize the artwork to exactly what you want. It's just a question of time and not a lot of it either.
>So why does everyone seem to think that art is only flat, lit, pixels on a flat surface?
Everyone doesn't think that. It just happens that digital art, both as a medium and as an industry, is the only kind of art relevant to conversations about AI, because that is the medium of artwork that AI generates, and that is the industry that is being disrupted by it.
Yes, that is interesting: so digital artists can go back to traditional methods. Like a game where you scan in crayon textures. It's a unique look. More handcrafted.
Maybe AI will bring about a sea change in how we express these traditional works in the digital context.
It's true we miss a huge chunk of art by just considering pixels. Museums are dedicated to the idea that art and expression are intertwined, and that context--both past and present--brings a unique experience to the observer.
Curators certainly aren't generating those longform descriptions next to art pieces. Someone had to think deeply, analyze, and type that out.
AI will make many, but not all, artists redundant.
As usual, the lower skill/value work will be automatized sooner. Platforms like fiverr will probably suffer a lot, but so will many "for hire" artists. High value work will remain to be exclusive for humans.
Many high value artists (programmers ...) started as low value and worked themselves up. But if the low value segment is eaten up by AI then the path will be destroyed as well. I wonder if this will make classical education more valuable (after graduation you're starting as high value producer) and further increase class division.
I don't think education has anything to do with this. AI tools are pretty easily accessible to anyone and you don't need a "classical education" to wield them.
If anything, I think the necessity of education is going to be weakened further and more quickly in the next decade.
> I wonder if this will make classical education more valuable (after graduation you're starting as high value producer)
I haven't noticed a correlation between formal education and the people that I would describe as "high value producers". That could be a bias because the median person without a degree doesn't get hired at all, but if we're removing mid-tier and below devs from the work pool, that's going to eliminate most of the people I know with CS or SE degrees.
> I haven't noticed a correlation between formal education and the people that I would describe as "high value producers".
I agree, but I would argue that people without formal education have to usually work their way up.
Basically, if I'm a high school drop out, then I'm very unlikely to land a job at FAANG right away. Quite possibly I will have to start in some low value job to get some credibility. But if AI eats most of these jobs, then it also removes the possibility for the high school drop outs to move up.
Only society can make artists redundant. If people want to get their "art" by digitally rehashing what real artists did in the past then we don't need artists.
How exactly things will play out is not clear: photography did not eliminate painting. Algorithms will not eliminate more "manual" creative work. Some new genres might emerge.
Ultimately what is a more important problem is that the commercialization of artistic production was always challenging. When you can create infinite replicas with semi-random variations it only makes the problem worse.
Exactly. Art as an economic industry may die out. As a career, it may be reduced to a much smaller scope. But we'll always be able to express ourselves artistically - maybe even more so if AI frees up more time in which to do so.
...or enjoy making art with other people to use in our commercial endeavors ?
Like, it might actually be fun to work with creatives to make art. It's one of the areas of my work I actually really enjoy. I can't imagine prompting DALL-E to be as fun and enjoyable, nor would it be "creative".
While people are (to varying degrees) generous, its hard to build everything on generosity and recirocity. On the other hand once you start bean counting and transactionaling everything and in particular not even accept different types of beans, out entire existence becomes the one-dimensional, money driven disaster it has become.
Many artists are really scared by what these apparently uncontrollable interests are unleashing. Who can blame them.
I wouldn't say thanks to information theory, but rather on human tendency to be bored with repeated styles. It doesn't matter how good the tech is, but if it comes out like Michael Bay movies or anything else that we learn to recognize, people are going to want 'more original' art. The other difference is that the point of art is to express and evoke emotions, it's an open question whether that can be done effectively without having feelings during the process.
This is a good take and under appreciated generally. Already theres an identifiable generic flavour to much AI art. Just knowing that it’s AI generated makes it feel somewhat lifeless. Not because of the end product is lacking , but because of the awareness of the production method. it’s less engaging if you know a machine spat it out, especially once you recognise and become bored by the stylistic markers that indicate machine generation.
That will play into the spending calculation of, for example, advertising agencies, particularly high end brand work.
I’ve seen some screenshots of ai ads already. They are usually somewhat horrying, like the person will be smiling, but you start to notice their mouths are open too large and they have too many teeth and lack eyelids, and the people in the background are even more distorted and nightmareish.
I wouldn't make that strong a statement anymore. The level of play by AlphaGo/Zero seems to demonstrate creativity and understanding that we would only attribute to humans. That is to say I can't say for certain that emotion couldn't be faked, and we don't know that human emotion is super-special, only that it's complex.
Of course it will make them redundant. All it takes is one "creative" in place of several graphic designers. What you write are just theories. Do you know what the design process looks like? How many "human problems" are there along the way?
Only the ARTISTS will remain. Artists whose work you will want to buy. Such a fancy for the rich.
I hire artists to turn my terrible sketches into decent looking stickers, generally in a fun, cartoony style.
For the last two ideas I tried midjouney and the best I could get from it was an additional reference to send to an artist. I watched a few YouTube tutorials that offered some interesting techniques for getting better and better images from midjouney, but to my eyes those midjouney experts seem awfully flexible about what the final output will be. They’re dancing with the AI but not fully wielding it.
Perhaps next time I’ll hire both an artist and a midjouney expert and compare results that way.
Look into ControlNet. Midjourney is not very customizable. Checkout reddit.com/r/StableDiffusion for examples of controlnet. You can turn your sketches into art with it.
That is a very good point: having tight control on style and composition is significantly harder than getting a good-looking image with those techniques (it is doable but it takes practice to learn how to engineer prompt accordingly)
Wish I could filter this conversation to those who are actually artists by profession. A little hard to hear a whole bunch of non-artists telling them what they should do or feel about this.
This is a strange line of thought. I mean, yes we'll get a whole bunch of bullcrap posts from people not involved, but effectively you're instead reducing the conversation to "Lets give the artist their 15 minutes of neo-luddism".
Artists are not the first group to fall under the hammer of technology and automation, and they will not be the last. I'm sure most blacksmiths thought themselves artists that worked hard to master a craft after years of work, and now a press stamps out that same work in seconds. This is hitting people hard now because things we think of as distinctly human are now being accomplished by machines (though was this not true in the past?).
There are discussions for artists here, but there are plenty more for all of society. Jobs will get replaced and change in form at an ever increasing rate due to technology if trends keep up. Will the rate of technology change job requirements faster than humans can retrain? If it does what are nations and societies going to do about this. Much like the AI safety issues, we need to answer these large scale issues now before artists and programmers are stabbing each other in the streets for breadcrumbs while multitrillionares that own the technology live like gods.
That's interesting. HN has a software bent on startups and such. As if programming is one of The Known Ways to create something. Fundamentally, our instructions are executed by a machine to some effect, one of which could be profit.
I wonder if artists are having their "software moment": a decade from now (or sooner) they will say, "Adobe Photoshop is the assembly language of digital art; use these AI frameworks/prompts/GPU cloud mix to do X and Y."
From that perspective, it's just pragmatic. No one writes assembly unless it's either for passion or practical.
I think that “AI” (there is not even a drop of an actual AI in the current technology) is overhyped and barely can handle any of jobs people wanted it to do. It’s maybe good for sketching, be that code, text or images, but not for final product, and won’t be any time soon. As it was with all “AI” products in the past, it’s over advertised by ML bros and grifters.
Half an hour trying to force one of the models to spit a manga-style picture of a character... with correct number of arms and legs. No successes so far. It's frankly disgusting most of the time, prob due to uncanny valley-like effect - it's kinda close, but not quite there. Not to mention, I'm sure I did not put "amputee" in the prompt, but I got a few pictures that should be honestly tagged with "guro"... and I wasn't even trying for NSFW!
Same experience with code: Github Copilot is 90% right 100% of the time. Its suggestions for comments and docstrings are so bad I would never accept them in a code review, basically your old "a++ # adds 1 to a"; the generated code is always wrong, and the bigger the chunk of code generated the more wrong it is. It's kind of OK as a replacement for Ctrl+C/Ctrl+V and for generating boilerplate... which shouldn't be there in good code anyway.
I suspect that for every impressive output we see there are tens of thousands of trash outputs someone had to wade through. Further, the less popular your prompt and the smaller the representation of what you want in the model, the harder it is to get something even remotely resembling what you want. Though, I'm not a "prompt engineer" (WTF is that...), so maybe I'm just using it wrong...
You guys need to understand, that the skill ceiling for AI art is very high. Its not some iphone-selfie tier technology.
Go to Civitai.com and huggingface, to start comprehending how many AI models there are, and how powerful the latest models are.
It takes at least two weeks of intense usage to get the hang of the basics/parameters and produce good output. AI art in the end will probably be dominated by professionals because of the increasing skill requirements.
Pray tell, what are you going to do the moment it is not overhyped?
Sitting around on our asses going "We don't have AGI/ASI yet, nothing to worry about" is kinda like saying 'nuclear bombs are nothing to worry about' in 1940 while watching your neighbor create an every larger pile of uranium. To think we're not going to accomplish this eventually is foolishness, and when we've already accomplished it, it's going to be very difficult to setup a legal framework that reigns in corporate interests around it.
What are you even going to do? We have no agency to do anything with the march of society and technology. Business interests rule and we are fed information that aligns our interests with those as much as possible. I mean climate change is a much more real evil than startrekian fantasies about ai, and yet no one here is doing anything or marching in the street or blowing up oil refineries. There should be ten threads a day on the environment here if people gave a crap like they should. That should give you a preview of the level of political engagement of the current residents of the earth.
I’m gonna be doing whatever I’m gonna be doing. Job market changes all the time, and people adapt. I feel pretty secured (at least for some time) for my field, which is a lot of R&N in wireless networks/radio and embedded systems.
My point is that current “AI” is just very well learned models, but it can’t really do intelligent work on its knowledge, can’t innovate, can’t infer, doesn’t have an actual thought process.
Founder of dreamphilic.com here. Based on our users I am seeing some trends:
1. Great artists are generating great prompts and thus great images because at the end it all comes down to your imagination.
2. I expected most users to be young population but what I see is they are mostly people of age 35 and above.
3. Creative people are enjoying it. People who are not good at using artistic tools or don't have access to those expensive tools but have good imagination can also see their imaginations come to life, thanks to AI.
I want to make a cartoon. Who can produce the imagery faster? Either spend $ commissioning an artist or spend money entering text in to a computer.
"A teddy bear walking though a park". If I don't choose the artist, the artist is out of the job and the AI wins.
You could say someone needs to train the AI with an artwork dataset but once that's initially done, its over, considering that studios are the main supplier to artists income.
Be realistic and practical here, in what situation are you going to start typing into a computer "A teddy bear walking in a park" and then start to do anything with the images it generates in a case where you would've entrusted and artist or designer to make something for you ?
What about when you need different scales, different colors, different this and that, are you going to sit there and keep prompting for this ?
I've worked with people who are really good with Illustrator, they can change and produce images for different purposes in a ridiculously short amount of time already. The ting is, even if I could recruit AI to do it, I probably wouldn't and just want someone else to take care of it. To make favicons and screw around with thumbnails and vector images and the list goes on.
There's also the fact that, some people have an eye for style and application of it. Even when I've tried to use DALL-E, which I never have successfully because it always generates something weird enough to be unusable, I've not been confident that my own personal selection is good enough, I'm not a designer, I don't have an eye for art or color, or style etc. It's not my profession to know what's nice. I don't want that responsibility, so I outsource it.
The problem arises if a bunch of studios start adopting AI assisted workflows to increase productivity and the studios that are holdouts get left behind because it would no longer be economically viable to continue animating without AI. It could be similar to how traditional cel animation has declined in use. If AI assisted output clears the bar for consumers, then it's probably good enough for managers.
People using AI professionally probably won't put the raw output of DALL-E straight into the finished work either, I've seen artists that use Stable Diffusion to generate a base image and then do heavy amounts of editing with Photoshop or similar.
AI won't make artists redundant because it does not experience human sensations. The only thing an AI can do is read the flow of data we make available to it, which is text, audio and video.
An AI will never be able to represent what it means to be subject to police brutality, because it cannot feel being beaten and humiliated by a cop.
An AI will never be able to paint an anti-war poster from the perspective of someone who has felt the ground shake due to shelling of their hometown, because an AI has not lived there for 30 years and does not know the feeling.
The only way AI will make artists redundant would be if it became completely human, which is contradictory since at that point it would be a real artist
In case you don't know, AI art is not an AI pumping out art by itself.
A human prompter has to prompt the AI.
The prompter can perfectly understand 'police brutality'/'shelled hometown', as they are human, just like the artist.
Artists feel threatened, because they invested 90% of their time in drawing skills, which went from unthinkable to automate, to AI surpassing 90% of artists in 8 months.
> they invested 90% of their time in drawing skills, which went from unthinkable to automate,
Except that the AI doesn't automate drawing, it automates the creation of an image of a drawing.
I have a friend who is an amateur water colour painter. I have no doubt that an AI/ML application can produce pictures that look very much like pictures of his pictures. But at the moment at least it is utterly unable to create the actual artwork.
All of that is true, but as soon as you model subjectivity and embodiment with any kind of credibility it's game over.
The current generation of tools doesn't do that. It's essentially a very sophisticated parrot making speech-like noises it doesn't understand.
But a couple of generations from now, I think it's going to be much less straightforward.
Also, the thing about genius is that it's subjective. In the arts it's more or less synonymous with mastery of a medium with impactful novel insight.
You could argue that AI systems are well on their way to mastering visual media. It's not quite true, but that's because the people training these systems are not artists, and so far they're selecting work that looks a bit Social Media and Game-Ish rather than Art Museum and Contemporary Show-ish.
I don't see any reason in principle why that couldn't be fixed with better training.
So what about impactful novel insight? The point here is there's a kind of cultural and perceptual feedback process which selects certain works out of semi-random cultural noise. There are always a lot of artists making a lot of work, and most of it is not that interesting. Selecting interesting work doesn't require intent, it just needs a feedback loop.
So if you create a situation in which a community rates the art and selects certain works/algorithms/training sets over others, I suspect the impactful and novel insight will happen automatically.
Sentience or subjectivity are not required. In fact you'll get an automated version of what happens already, where different communities with different levels of education and sophistication select different kinds of work for their own reasons, and some are considered "works of genius" for reasons that may be as political and cultural as artistic.
It's not that clear-cut. Humans can empathize, to the point where observing someone getting hurt flares similar neural response to the one in the hurt person. Sufficiently vivid imagination in conjunction with a lot of research can give you an experience "close enough" to the real deal to be able to write about it. I've witnessed an author falling into a deep, clinical depression solely due to the subject they decided to tackle. Not every person is capable of such a deep dive into an experience they nominally don't have, but authors and artists tend to be able to do this. In such a setting, "personal experience" is a fluid term, not necessarily synonymous with "he was there at the time personally" or "it happened to him personally".
When you wake up from a nightmare, you're covered in cold sweat, you have trouble breathing normally, your hands are shaking - did you "personally experience" what you've dreamed of? Authors and artists are in a business of dreaming like that while awake, and sharing those dreams with others.
The AI will surely get there at some point, as others noted, once you model embodiment and imagination it's game over. But it's not as close as others seem to think, in my opinion.
You’re being downvoted for contradicting the AI hype machine, but you are completely correct. Humans have been creating art for at least tens of thousands of years, and none of the reasons we create art will ever be fulfilled by AI.
AI just needs to be better at creating art from instructions than a hired artist is.
And we already know great art can be created following a prompt - hired artists interpret their client's instructions all the time (concept art, game art, movie art, etc.).
hired artists interpret their client's instructions all the time (concept art, game art, movie art, etc.).
I cannot disagree with this comment any more, I hire creative people to help me to create things which isn't in my skillset. I don't hire them to tell them what I want entirely. Sometimes I'd at least describe what I want, but often they will say, well that's good but how about...
That is a huge part of the value add for me personally. If I knew what I wanted, half the battle would be one, AI or not.
I don't think we disagree. Artists help you turn a relatively short and vague description into a finished artwork. The clients usually already provide very little information in comparison to the number of decisions needed to complete the work (e.g. choosing composition, theme, style, colors, lighting, etc.).
Working from vague prompts is exactly what the current AIs do. Follow up questions and ability to do precise edits are likely next steps in the evolution.
I think artists will just have to leverage the tech on a larger scale to be effective. No matter how it's created, the best art will still have value..
It may not be economical to, for instance, have an entire studio of dozens or hundreds of artists produce an animated film with an anticipatory budget.. but I suspect it will still be quite profitable for a handful of artists and some AI to produce that same film in a shorter time period.
Or it may be profitable for a 13 year old kid to produce that film in his spare time. In that way, I see it as a sort of artistic renaissance where new voices are found and the pace of artistic acceleration is unprecedented.
I wait when game from like Dark Souls series is released and all concept art is generated by AI, to me it seems such game will have no style consistency, it will look like game that is is put together from market place props (eg. cheap garbage).
Because AI can generate very good looking art works, I don't think it is all there in making compelling, consistent, and visually pleasing art, simply because there are feed back loops, between multiple people, from concept art to end product.
What's sad about AI is peoples reaction to it. The defensiveness from artists misses the point: Technology should better our lives and free us to pursue our passions. The real problem is under our current economic model, it has the opposite effect. Instead of getting defensive, we should be rethinking the role of our economy. Sadly (and cynically) I suspect we won't see change until AI comes for the CEO's and politicians.
>an open-source Stable Diffusion, which is a basis for multiple projects, including commercial Midjourney.
source? The model behind Midjourney looks quite different from Stable Diffusion, to me it looks like a model conditioned to image embeddings (like Karlo or Dalle 2) instead of text embeddings, also I don't see any mention of the Creative ML OpenRAIL-M license on their website, when they experimented with SD in August they added it but it has not been there for months, To me it is clear that MJ has trained a new diffusion model from scratch, it is also noticeable by the difference in performance, even for a model trained in AI image recognition it struggles with images generated by Nijijourney (MJ but finetuned for anime) while it has an easy time with anime images generated by SD and its finetuned models [1], which means that the distribution of images is very different.
I think so too. AI might affect applications like Photoshop if the result of the provided output is good enough. Stable Diffusion is going into this direction.
While ChatGPT is nice, real authors and journalists create value by bringing in a nuanced and subjective insights based on their research/expertise/ideology.
Personally, I highly value the vanilla perspective of chatGPT's explanations compared to picking through some authors bias to try to find truth for myself. What you are calling value is nice for entertainment purposes or perhaps some heuristic understanding and essentialization... But it is usually at expense of nuance and truth.
It doesn't matter whether diffusion programs or even more rarefied approaches to solution spaces can capture the essence of human-produced art.
Markets are fundamentally about perceived value instead of actual value. Algorithmically generated creative works do not have to compete in the field of actual value; they has to compete in the field of perceived value. Let's be real, almost all customers of art - whether they pay before or after the production - are pretty bad at estimating actual value.
More philosophically, it's not possible to empower humans above human superorganisms. You make a successful walled garden or you fail to reproduce your ideals into the future.
That may be true for graphic art for movies, video games, websites, etc, but not for fine art. In fine art, the resulting artwork is judged much less by the result than what went into it. The artworld already is deluged by art, so the narrative around the art and the artist is the most important differentiating element.
I have a hard time being worried until AI actually becomes comparable to human intelligence (and then art will probably be a lesser concern amongst others). Why? Because art is about the generation of things that people find appealing, which you can automate, but it's as much about am emotional conversation between people (using a "language" words can't replicate), interest in artists as people, and doing novel things. My suspicion is that whatever great thing people can do with AI, in the ways it exists now, people will get bored with it very, very quickly and move on to something they feel is more soulful.
What I've seen from all the AI lately is a lot of vague uncanny valley images that somehow don't fit. They're too perfect and even when they try to be imperfect it's too intentional.
AI won’t make artists redundant because you don’t pay an artist for a picture in the first place, you pay them for their creativity, their ideas, which may eventually result in a picture. You know, their artistry. Too many artists devalue themselves and their livelihood by defining their work as entirely mechanical, entirely defined by their material product. Art is more than just produce. You’d think artists would be the first to say that.
Imagine a world where AI art tools are commonly used.
Ignore the messy legal and moral issues; just imagine, for a moment. What's changed?
These tools effectively enable artists to:
- generate a variety of concepts for some desired output
- render out the detail, lighting and shading that has traditionally been done by hand
- convert existing images and concepts into artistic 'styles'
You want a set of icons? You want a custom font?
The effort is reduced from weeks or work to days, if that.
People will have to learn to use them; just like any other tool.
So really, what actually changes?
Two things, specifically spring to mind:
- The existing skills that people have to do these things become obsolete.
- The number of people needed to do the current job that artists are employed to do is drastically reduced.
That's what's going to happen, and it's happening already. You need people to paint tween frames for your animated video? Well, 2 guys in a basement can do it now instead of a team of 10.
Yeah, you still need the key frames... but the mechanical shading and drawing that employs of lot of people, specifically in the animation industry, is going to be ERADICATED.
In corporate teams, will your 'design team' of 5 people be cut to 1? Probably not.
...but, those industries don't employ the majority of creatives. The majority of artists do not draw concept art. They do mechanical technical processes in VFX and animation.
There will be new companies, and new roles.
...but, a lot of people will find the mechanical work that earns them their daily living wage will replaced by a much more efficient automated process. Those people may, perhaps, be able to re-train and get new creative roles doing other things.
...
What the 'anti-AI' movement has won is a short reprieve until the 'ethical' and 'watermarked' corporate AI art generators roll out, and get accepted as industry standard. Maybe that's a good thing? I dunno.
I think it's fair to sat that the cat is out of the box now.
In the future, there will not be industrial scale human drawn 2d art. That industry will no longer exist. Hopefully the people currently in that industry can find other creative roles (or retire) by the time that happens.
When photography came out, it had a similar impact on hand drawn photo-realistic paintings; and I think you can see thing, historically. It's not that no one does it any more. It's that it has become a niche job, that is technically inferior to just taking a photo.
I don't believe they'll be successful in stopping things from happening... but I think it's pretty undeniable that it's caused things to slow down in wide scale corporate adoption.
it sure as hell will displace artists and wreck art market. people are already choosing to pay couple bucks for an app to get some "art" from a random company that repackages stable diffusion, which was built on unlicensed art, instead of paying an actual artist to create something. and when called out on their choice, people make excuses like 'it's just for fun', 'it's cheaper than paying an artist', 'i don't have money for an actual commission' (and yet, somehow they have enough money to pay for an app). apps are currently actively undercutting and pricing out artists, and it is working for the apps, which also get free promo with every customer that gleefully posts their results.
apps can set any price they want on what costs very little to create - there's some compute, storage, etc. - but they can just repackage existing models, train on unlicensed artwork, get all the parts for free, and then price it (and undercut) any way they want. there's a lot of leeway on how aggressively can a service like that price it's "goods" which are basically free to make, and can be made in no time, at enormous scale. there's infinite possibilities for predatory pricing when the base price is approaching zero. actual human artists, who do commission work, which take time to complete, and are worked on one piece at a time, can hardly compete. even an entire market (or a piece of it), full of different artists, hardly can.
That's only a market problem if they are "dumping", i.e. selling under their price of production in order to gain more market share. Like Uber subsidizing their actual product with VC money and gaining more market share. I think you can sustainably sell the images for pennies as long as you run your own hardware and keep a lookout for unnecessary costs/middlemen (AWS, various paid Google products, etc.)
Apparently, undercutting (and other forms of predatory pricing) is not a market problem.
To reiterate, they're pricing out other artists, and undercutting artwork (such as, commission work) prices, not just the price/cost of generated output (which services can also undercut/predatorily price between their own kind, going as low as 'free' - 'but you'll have to pay us extra for priority/compute/features/whatever').
An artist might be offering commissions for, say, $10. Someone on fiver might do something for $5.
An app comes in, and offers you a package of 20 "commissions" for $2. See how that undercuts artists? It's frowned upon even within artist/trade circles when artists do that sort of thing (as overt lowering of prices can affect everybody else in the market), let alone when a random non-entity comes into the market to disrupt it (and that is the intent, when such services specifically advertise their 'art-making' ability), and does so while operating on your own unlicensed (stolen) artwork.
This entire article misses critical developments in the AI art space. Controlnet was just released last week. It allows for precise fine-tuning of the image with skeletal positions, depth maps, outline sketches, etc. It has been exceptionally received in the AI art community, because pure-text-prompting precisely runs into the issue described in the article. You can only convey so much information with text prompts.
That being said, artists in general still don't want to touch controlnet. Because despite controlnet solving their complaints about how the AI isn't controllable precisely, it doesn't solve the real problem, that drawing skills are massive devalued.
Artists will still exist, but most likely as hybrid 3d-modellers, AI modelers (Not full programmers, but able to fine-tune models with online guides and setups, can read basic python), and storytellers (like manga artists). It'll be a higher-pay, higher-prestige, higher-skill-requirement job than before. And all those artists who devoted their lives to draw better, find this to be an incredibly brutal adjustment.
PS: Despite how people made fun of 'proompters', or predicted that prompting would be automated away. The skill ceiling to good AI art has radically increased. There's now 10 different fine-tuned models you need to learn the basics of, each with different strengths. There's thousands of LORAs to insert into prompts to precisely reproduce a subject that the base model has no information on. There's 20 parameters to tune, each with different effects. There's VAEs which affect coloring and fine-details. Now there's controlnet, and you'd better learn blender to rig basic skeletons to feed it. This likely suggests the future of an AI-augmented economy: People become AI wranglers. Wordpress was supposed to make blogging easier, instead it spawned an industry of WP wranglers.
> that drawing skills are massive devalued.
This is the critical insight. Human artists having nothing to worry about. Human drawers do.
The decoupling of concept from execution is old news in most forms of art. Writing music and playing an instrument are two very different skills, and indeed any session musician will tell you that playing instruments well is not as valued as writing amazing music.
In fact we went through a similar thing in pop music in the 70’s and 80’s, with the brief moral panic over electronic music and people who “just push a button and don’t even know know to play their instrument.” Turns out nobody cares, as long as the music is good.
AI art decouples concept from execution for visual art. That’s all.
You're not taking into consideration how the actual artists actually feel. And, from the first hand - the feeling is horrible. Most artists love to draw, the process itself is incredibly satisfying. Just like a musician loves his instrument.
This is destroying them, stealing the joy from the thing they devoted decades of life. My wife is extremely depressed by this, to the point I think she'll need serious therapy. She still has most of her job, but yeah recent developments like ControlNet? Well, shit.
59 replies →
"This is the critical insight. Human artists having nothing to worry about. Human drawers do."
The reason why amazing music is valued so much is because there's so much music that you have to be amazing to be noticed. Art has had that problem for a long time. Someone spends 100hrs of talent on a masterpiece and we say "meh, seen a thousand of that quality"
Learning theory is easier than applying it so becoming an artist is easier and now artists have more competition. Not advocating against AI art but it's obviously going to have a negative effect.
3 replies →
Even human drawers might do all right, if they're selling physical drawings. People will still put hand-painted canvasses on their walls, just like people still enjoy an acoustic guitar concert, even if the guitarist isn't world-class. The authentic connection with a human being is where the art survives. If you're not an amazing artist, you can still be an authentic one.
But if you're just churning out commercial illustrations, then sure, an AI can do that now, or soon.
> Turns out nobody cares, as long as the music is good.
Actually, all those people who lost their professions they spent their lifetime working on, they care, rather a lot.
> indeed any session musician will tell you that playing instruments well is not as valued as writing amazing music.
Yes, in 2023 they say this, but I assure you when "session musician" was a common profession, that no one said this.
When I first started in music, if you wanted your piece of music, you had to pay instrumentalists to play it, and this was a valuable professional skill.
> Turns out nobody cares, as long as the music is good.
And the music is not good either.
When did everyone become so sociopathically detached from the welfare of others? It's just horrifying.
1 reply →
I'm not an artist, but my fiance is an animator. Generally, something similar exists in stock assets (photos, assets like vector files, other such things). We had a discussion on whether the existence of stock assets means drawing is no longer necessary to be an artist, and her answer was it still is something you should be able to do because knowing how to draw is the best way to learn the fundamentals of art in general, knowing how parts of a body work in a drawing or animation, and generally something is "good" vs. "bad," etc. However, as someone who isn't a fine artist, she doesn't have time to generate everything frame by frame, so she doesn't do it for every project she works on, but that knowledge is invaluable and one you can only obtain from knowing how to draw.
I think the best analogy I understand it is as is it is like assembly language or low level programming in general. I certainly do not have the time to program every piece of code I write in asm but having done projects in asm is invaluable as a coder[0], given how much it informs my mental model of how the code I write actually works. That understanding is beyond valuable and is something that puts you a rung above everyone else who just copies things from SO without knowing what they do. I think AI for devs, to the extent that it will evolve, will still be as such. People who primarily find it amazing today I find either 1) use it as a productivity boost for things that need a lot of boiler-plate[1], or 2) are SO-copy-pasters who are just amazed they have to think even less. A LOT of the "AI artist" community are the artist equivalent of the 2nd, honestly, and it's easy to detect AI art because it's generated by people who are not really artists, and in similar fashion either don't know the fundamentals or who only create "good" work by nearly directly copying other art pieces.
Also, to critique your analogy, what you guys are saying is along the lines of "given DAWs, why learn to play an instrument at all?" Except a plain look at any good composer will show they know how to play at least one instrument, even if they don't play all their music on that instrument, there is no doubt knowing how to play clearly makes you a better composer. I mean, can you even imagine a composer who cannot even play piano, or guitar? Sure a composer need not be a virtuoso concert pianist, but they should at the very least be able to play the chords of the very song they've composed. Nothing thus far teaches the human mind deep understanding of something more than doing that thing does. It is so clearly obvious in music and programming and the only reason people keep saying you can be an artist having never learned to draw is because such people simply do not understand art or composition at all.
[0] I'm not really a software developer but a computational scientist, and so I won't say I'm a developer, but still understanding of data structures, algorithms, discrete math in general, and yes asm is invaluable for me.
[1] As someone outside of the dev space, I don't know why you people don't just make better interfaces instead of juggling 9 yamls and 3 environments for every project, or just roll your own interface and reduce the boiler-plate yourself.
> Because despite controlnet solving their complaints about how the AI isn't controllable precisely, it doesn't solve the real problem, that drawing skills are massive devalued
At this point if I was an artist I'd probably be looking for a way to leverage my existing skills into AI art, because it seems like drawing skills will go the way of the horse and carriage.
I feel bad because this has happened so quickly. Previously it seems like there was a longer transition period.
This kind of transition has got me thinking the same about coding though. If AI programming does indeed take over, my programming skills will likely be massively devalued.
If you believe this is going to happen, how should you prepare for it? Start learning prompting now? Get involved with building these AIs? Etc.
I've tried to convince artists to transition to the AI-augmented future, since October. There's almost no successes. I've seen artist communities on multiple websites, in multiple languages. At first they tried to laugh at the AI. Now the AI has improved so radically so fast, they universally prefer to stick their head in the sand, and ban discussion of AI altogether, its all denial and rage.
On the more optimistic side. I see an extraordinary explosion in artistic innovation, just look at websites like CivitAI. The massive community all training subcomponents of the models for each other to share. The models rapidly improving every month just through fine-tuning and theoretical innovation, without stabilityAI's involvement (They are distracted by lawsuits now). There are many 3d-artists intensely experimenting with AI art, to say make AI-anime, which has illustration qualities on every frame (A previous impossibility due to the costs involved).
It seems with AI, it'll really cleave communities in two. The ones who eagerly embrace it, seem to enjoy it extraordinarily, and achieve quite a lot of popularity and success. But the rest just want to pretend it doesn't exist, waiting till employers realize that they are no longer needed.
Regarding programming, it doesn't appear that AI programming can replace humans. Programming is very similar to novel writing in terms of complexity for AIs. And AIs are still extremely terrible at long-form storytelling. The lesson is to aggressively use AI tools as much as possible, to understand the long-term weaknesses of AIs, and deliver your values in those areas as a human.
10 replies →
> seems like drawing skills will go the way of the horse and carriage.
The exact same thing happened to technical drafters when CAD destroyed the entire (sub-)industry.
And to typesetting, for that matter[1].
Which is a pity in the human sense and the sense that both were forms of artistry and produced things of great beauty, both the product and the machines used to enable them, but they're simply not economical in the face of Solidworks and digital composition. And yet the replacement technologies have also enabled a lot more creativity and further advances. Objects with complex geometries are now possible to specify and manufacture, when they previously could not even be accurately drawn.
1: https://en.wikipedia.org/wiki/Wapping_dispute, where not only could the unions not simply oppose the rising sea-level of technology, but their failure seriously damaged union credibility in general.
> At this point if I was an artist I'd probably be looking for a way to leverage my existing skills into AI art, because it seems like drawing skills will go the way of the horse and carriage
Drawing skills were made redundant for producing most types of high quality image before the motor car but people still pay for hand drawn items
Not sure AI lowering the skill barrier for and speeding up the generation of digital art is really going to change that
> my programming skills will likely be massively devalued
I see this in the context of the drawing vs. art analogy. Yes, your typing and maybe syntax skills will be devalued, but your higher level, creative programming skills are probably safe for a good long time yet.
13 replies →
>If you believe this is going to happen, how should you prepare for it? Start learning prompting now? Get involved with building these AIs? Etc.
If you still want to work as someone who produces code - except your personal code factory has changed from brain and fingers to AI - then yeah, you should probably do both of those things.
Even if you spend 1-2 years and this AI hype doesn't work out, you still have those skills to go back to.
Frankly, this mainstream adoption by Google and Microsoft is, uh, not going great. So you can afford to observe for now, but AI advancements have been made very rapidly, so it's not wise to completely ignore it either.
> If you believe this is going to happen, how should you prepare for it? Start learning prompting now? Get involved with building these AIs? Etc.
Short of advocating for artists' unions...
Law is still going to govern anything AI creates, so anyone who develops a skillset in both art and law would be well-positioned to gatekeep as a copyright troll. This carves out a career for yourself and enacts revenge by making it a liability for employers to use AI to replace artists.
HR is the theocratic version of that, since law school isn't cheap. You can still gatekeep, but you'd be playing by more-arbitrary rules ("you can't use that AI because it incorporates images of Women Without Penises," etc.).
Security is pragmatic, but I worry about the long-term stability of it. Every time a breach is announced, there are no consequences, so why even pretend to need it?
I used to suggest pivoting to tangible works and experiences (architecture, sculpture, etc.), but those are easily displaced as well-- I would bet against it now.
Look at porn-- what started as a handful of performers in studios is now done by any college student with a webcam in their bedroom (next up to be replaced with AI-generated content).
For now, you can still add value as a sex worker by offering the GFE, but even that's about to be obsolete. You can get your fix of flirting from a chatbot (TTS or sexting) and 3D print a copy of anybody else's genitalia. Remote-controls, vibrators and fluid pumps add some life to it. And you don't even have to leave the house, which affords you privacy to pursue darker subjects without oversight and save you more time for repeat consumption.
Truly, what a wonderful world...
Go with the flow. AI will need humans to be effective, humans will need AI to remain competitive. But everyone will have the same base models, just like we all have the same web search and electricity. AI won't be a competitive advantage, it will be a basic requirement.
Do you believe the number and complexity of software applications will decrease in the next 10 years because of AI, or that it will spawn whole new ecosystems of software and new types of jobs? I believe the second is more reasonable, we will have higher expectations from software in 2033 than in 2023. The easier AI makes it, the more difficult we make the tasks.
Human desires fill the available space like air, AI exponential is slower than our entitlement. So we still need to work.
> If you believe this is going to happen
I believe this is going to happen (and is happening right now, in front of us) not just for programmers but for any role that can be automated.
> How should you prepare for it?
Get equity. I mean ownership stake. If you don't have a legal/financial claim on the output of the machine then you're about to be part of the worthless surplus from the POV of the system.
> Start learning prompting now?
No, the machines will do that well enough in a minute or two.
> Get involved with building these AIs?
No, the machines will do that well enough in a minute or two too.
- - - -
Artificial Intelligence destroys scarcity, which is the fundamental basis of our societies and economies.
Think about it: scarcity is the very problem that societies and economies evolved to solve in the first place.
Now science and capitalism have delivered technology and wealth. There is enough to go around if we just worked out the logistics, and computers can do that for us in a matter of moments. In other words, the "World Game" is not hard! https://en.wikipedia.org/wiki/World_Game we just have to get over our hangups.
Now here's where it gets really interesting: ChatGPT et. al. don't have glands, they don't have emotional trauma, no PTSD from being humans-on-Earth for generations, etc. We can program them to be sane and perhaps even wise.
We can also attach empirical feedback devices to them, make them scientists...
So we have physical abundance and benevolent, sane, empirically-grounded AI advisors, how much longer will it take to sort things out? I think we could be looking at the start of a Golden Age?
Yeah this post seems like it will not stand the rest of time.
The fundamental conceptual shift is the emergent behaviors displayed by these models. And that is highly unforeseeable.
We’re not working with machines that do what they’re built to do. They’re doing more. And every day we discover something new they’re capable of doing.
Making any sort of statement about the limits of this technology is going to be shortsighted.
The opposite is also true. People overestimate what is possible and underestimate how much more effort and time is required to start replacing industries.
> Artists will still exist, but most likely as hybrid 3d-modellers, AI modelers (Not full programmers, but able to fine-tune models with online guides and setups, can read basic python)
Why should artists have to learn fucking Python!? There are so few jobs open to people who don't want to "learn to code", this was one of the last, and that's going too.
> This likely suggests the future of an AI-augmented economy: People become AI wranglers.
So we don't get to create ourselves anymore. No drawing, no making music, no programming even. We just give prompts to AI programs. Which means we are worthless, because any bozo can do this.
I hate this timeline so much. It's like all the worst and most mediocre parts from every SF future, with an extra dose of stupidity and greed on top.
(And I'm a computer programmer.)
I wouldn't worry too much. We've pushed the biosphere past the point of repair. This "utopia" they dream of won't have a chance to come to fruition before we all go down in flames.
And given this dichotomy as choice, I would choose the flames too.
At the end of the day, we are not going to see AI art in the Guggenheim, or any of those other museums where the finest art in the world resides. AI art is and will always be a novelty. Real art comes from the human spirit. It is not about technical execution, and not just about painting a pretty picture. In actual art we want to see humanity, something that AI will never, ever truly know about.
Seems like a lot of people are missing this point, and it seems like a lot of people have a chip on their shoulder about being unable to create art, and believing they can express the deep recesses of their feelings and psyche and everything else that makes us human by proxy through AI, and thus completely missing the point of creating art in the first place.
These are great platitudes, and as a fellow human I like the idea that only we can create "real" art, but I don't know how to justify that idea. I have no chip on my shoulder about art, it's not part of my identity at all. Computer programming is, though, and I fully expect it to affect my industry in the next decade.
The current fine art market is extremely unmeritorious so I agree that AI art will never make it to the Guggenheim. That isn't a good metric since the human curators will never allow it to happen. The metric will be the industry. Will ai start composing scores for movies? Promotional posters? Art that's sold outside of expensive curated galleries (etsy, ect)?
It's possible that humans will keep their stranglehold on these and I hope that's the case... But I wouldn't bet on it.
5 replies →
I think even without AI art we are well past the point of art in famous galleries making sense. Isn't most of this just thinly veiled money laundering anyway?
I imagine we WILL see AI art in a famous gallery for exactly this reason. Except it will be lauded for the imagination of the creator and the tweaks they made, not their drawing ability.
1 reply →
> This likely suggests the future of an AI augmented economy: People become AI wranglers. Wordpress was supposed to make blogging easier, instead it spawned an industry of WP wranglers.
Good counter for "AI will steal our jobs". When competition starts to use AI+human, you got to level up. And since everyone has the same base AIs, the differentiating factor is still human.
thanks was looking for something like this.
SD has been great at giving lots of new ideas but frustratingly difficult/impossible to iterate on anything.
Artist here: Article is an idealistic take not based in reality- the Jeff Koons of the world will be safe with their 30 million balloon art sales.
Everyone else is going to be rekt.
All of the work my artist friends do is for businesses- businesses trying to make the biggest profit. Ai art generators are faster/cheaper/more flexible. The artists are done and now it's all about the prompt engineers that don't need to be artists to excel.
Market will decide what to pay "artists" but seems like it will be a race to the bottom. Artist types I know are preparing accordingly.
But you should realize that any artist learning prompt engineering should be a better AI-artist than a non-artist learning prompt engineering. After all, art is not jsut about technique but also creativity, composition, theme etc., right? And an artist can furthermore touch up the AI generated product manually if necessary.
Yep. I spent a good hour trying to generate some stuff with DALLE. I then asked a friend that works in media and he was able to give me ideas for a prompt that worked almost immediately.
AI art is fun and _sometimes_ it can replace some things but at the end of the day, I don't really want to mess around prompting over and over. If I pay an artist $100 for something, I couldn't care less how they make it. I feel like AI can be another tool. I'll even make a bold prediction. In 10 years, we will still have "Graphic designers" and "digital artists" the ceiling will just be raised similar to how it was when photoshop became more commonplace.
3 replies →
What leads you to believe those problems won't be solved in the next few years? What stops me from having ChatGPT provide feedback on my prompts to make them more artistic? Maybe I just pay some artist to touch my stuff up for me if they're desperate for work or maybe the output gets good enough that I don't.
1 reply →
Sounds like a job for us pot smoking slackers who studied art history to me.
Out of interest, how are they preparing accordingly? It's not remotely my own field but I'm starting to wonder how I should be guiding my children who have tallent for art, in light of all this.
Draw comics, or really manga in particular. The only real challenge in AI art, that won't be solved any time soon, is storytelling: chaining multiple panels together for coherent stories. Even ChatGPT cannot tell a coherent story longer than 2 paragraphs. Being able to write a long story well, requires understanding world modelling, human motivations , etc, AGI tier abilities.
Also, expose them to AI-art. If they lose interest in art after seeing AI art, that means they were never meant to be artists, they merely like to draw, not to make art. And drawing alone is not really economically useful anymore.
9 replies →
AI is shifting the balance of power from technical ability and quality of tools to creativity/vision (and ability to market). The only "art" job left is going to be something akin to an art director.
Expose your children to a wide variety of art, books, music, food, travel and so forth. Teach them why (aesthetically) good things are good and bad things are bad. Also, encourage them to create stuff for an audience so they learn how to present things, gauge people's tastes and become comfortable with failure young.
The harsh reality as far as I can tell is that the prospect of making a living as an artist, which was already slim if we are being honest, has now shrunk to near zero.
They should be learning about the AI field in order to create their art. That might be the only skill worth money when they are older.
Focusing on learning how to leverage these new tools to produce their own works faster/better/cheaper- focusing on their original ideas/designs/models- coming up with strategies to protect their source art from being used in other people stable diffusion models- learning about/utilizing nft tech to find new pathways to monetize their art-
In the end though its going to be all about original ideas imo- so creatives that don't have original IP really need to get on that and develop it- the new tools will allow anyone to create anything in realtime- people will be frozen by the prompt if they don't have original ideas/concepts/characters/worlds etc-
The temptation for those without original ideas will be to leverage chatGpt etc for "ideas" but humans are not required on that path-
So for your children- ignore the tools/technicalities and focus on ideas- original ideas-
And older artists said the same about computer art, things like :" this new artists do not know to mix paints or prepare a canvas". Artists can adapt and use this new tools and do their job faster, you might have to sell a logo cheaper but you might be able to make them 20 times faster.
There will just be people who are employed to wield these tools, select, curate and combine these images into whoever commissioned the work.
I don't know who all these tools are really aimed at? As a software engineer, I'd still prefer to pay someone to mess with any type of "art" or design while I focus on other things.
I mean even if I had an AI program that I could use to code, to do SRE, to do images, to do accounting etc, I still think someone would need to be in charge or making these things happen, or else I'd just be busy prompting machines all day, which sounds mad fatiguing and boring.
I think the real question is what does "making these things happen" realistically pay, pennies on the hour to sit in front of a screen and maybe elevate that to the 1/10,000 person who has to come hit a few keys to get the station running again?
1 reply →
On the other hand, where was the backstop previously? Companies could have saved money on art by hiring unpaid college interns or teams from where the median wage is less than a few hundred dollars a month. Evidently there was some reason to favor hiring actual artists versus the cheapest possible person who can draw. Maybe that reason hasn’t died out yet.
LOL https://www.nytimes.com/2023/02/18/arts/jeff-koons-sculpture...
My point is that unique digital art won't be replaced even by a hypothetical perfect AI image generator.
When it comes to the digital art workforce - it is likely to be strongly affected.
> Others might fall into a gap of “nice skills, but not yet that offer a business advantage”. Furthermore, the lower entry barrier to create any art is likely to result in the average quality going down - not unlike that plastic made manufacturing cheaper, but also less durable.
And you are right that, sadly, in many kinds of digital art, it will be a race to the bottom.
There's a mistaken assumption here. The author seems to believe that these generators operate on the basis of random chance, hence a 200x200 pixel image in RGB has a possibility space of 10^300000 images.
...But an overwhelming majority of those images are random noise, and are effectively homogeneous and interchangeable. They're perfectly identical.
Like human artists, the image generators are (very strongly) biased to create "structured" images, which are derivative of existing artistic works or natural representations in the form of photographs. In principle, there's no a priori qualitative difference between the AIs and humans in this respect -- no human art is created de novo, but is always a continuation of, or a reaction to, existing forms.
Further, there's no reason to believe that (a) the size of the possibility space matters in a quantitative sense, and (b) that AIs will never be as capable of qualitative "originality" -- in conception, composition, or subject matter representation -- as a human artist would be.
I put my caveats on indistinguishable images. And three is the challange - to make a better estimation of an effective number of distinguishable images.
> (b) that AIs will never be as capable of qualitative "originality" -- in conception, composition, or subject matter representation -- as a human artist would be.
Sure, AI already creates original things. The point is that questions like "create a drawing that presents friendship" can be approached from many, many different, and what is crucial - subjective, ways.
They are not worried about making all artists redundant, they are worried about making most artist redundant.
Instead of needing to pay for mediocre artist to get mediocre art that's good enough for a purpose, now the AI taught on the good artist (all subjective of course) can produce similarly mediocre art that's fit for purpose.
It's equivalent to replacing all of the simple CRUD web devs with AI that does good enough job
Interestingly, it might be easier to replace mid-skill artists than mid-skill developers (not to say both won't happen over time). The key difference is that when generating art, you can say "Make A in-the-style B.", get what you want, and move on.
Meanwhile, you have to live with the consequences of AI generated code.
Case in point: My wife was trying to build a system to generate POs per vendor from a series of workorders a couple weeks ago. She was using AirTable, which vastly simplifies the creation and use of relational databases. Given a lack of database experience, the structure that she made was misaligned with what she actually needed (and completely denormalized). Putting it into production would have caused major problems over time.
This made me realize that writing code is only one of the barriers to entry in software. Eliminating this barrier effectively gives everyone in the world access to a small team of incredibly junior developers who have no idea what they're doing.
Just like mismanaged DIY home projects create more work for professional contractors, I think AI might -oddly enough- create more positions for mid-tier and higher developers.
It most definitely is for simple reason - non interactivity. Image doesn't need to be told what clicking this button does or how that pop up menu needs to animate or million other interactions.
> Just like mismanaged DIY home projects create more work for professional contractors, I think AI might -oddly enough- create more positions for mid-tier and higher developers.
It might also create a lot of jobs for "AI crafters" - know enough about it to craft good prompts about what client wants and have enough graphic editing knowledge to tweak and mix AI input to create what is needed. It probably will also empower "one man shops" dealing in small customized websites for customers, replacing or augmenting what now stock images are used for.
Reading the comments it seems that everyone assumes that art is always and only presented as pixels on a screen. Whereas most of the art in my house is hand painted on canvas and wood, hand embroidered, hand knitted, hand woven, hand thrown and decorated pottery. One picture was hammered out of a sheet of pewter.
One of my favourite places is the Henie-Onstadt Art Centre sculpture park, another is Vigelands Anlegg in Oslo, no pixels in either place.
So why does everyone seem to think that art is only flat, lit, pixels on a flat surface?
I think the exact same thing, obviously because we're on hacker news where most people seem to be hoping they can spend their days prompting machines for art so they can replace "artists" and save some money.
It's actually funny when I think about it because most people wouldn't know what image to use and how to use it even when it was generated for them, there's even a skill in selecting art.
You could argue enough good photos have already been taken that all practically all photographers should have already been made redundant since the year 2000, we still have photographers.
I think if people are ignoring AI's limitations, it's more out of fear that programming is next than having something to gain. I doubt having the money/labour that's currently used creating art for something else would help me in a noticeable way:
-the money would just trickle up
-not that much money is spent on it anyway
-world becomes more depressing, nothing you look at had any effort put into it, no-one had to believe in an advert on any level, instead it's the output of a machine optimised to trick you (as one of a shrinking number of people with any agency) to spend and therefore make the machine stronger.
-artists/potential future artists decide to learn to code instead?
That reminds me of Ira Glass' quote about taste: artists know it when they see it.
Or why it's a mystery some folks prefer tabs vs spaces, or recoil from what they consider shoddy code.
Somewhere in the discipline of point, line, and perspective, in the composition of shapes and their organization, is the artifact of amalgamated neurons, to be observed by yet another consciousness.
I wonder if AI art will just help generate more gacha games. And one may wonder, of the limited time left on this planet, what really could we spend it on?
The library of every book contains no meaningful work in the search. The space of all generated art is oblivion.
Because that's what 99% of the people employed for their artistic skills are doing.
I doubt AI will have much effect on the tiny number of people who can make a living by producing art to hang up in your house. It's the people doing illustrations for magazines/websites/packaging/etc. who are fucked in the medium/long term.
People will 100% start hanging AI art on their walls. The killer feature is being to customize the artwork to exactly what you want. It's just a question of time and not a lot of it either.
4 replies →
>So why does everyone seem to think that art is only flat, lit, pixels on a flat surface?
Everyone doesn't think that. It just happens that digital art, both as a medium and as an industry, is the only kind of art relevant to conversations about AI, because that is the medium of artwork that AI generates, and that is the industry that is being disrupted by it.
Yes, that is interesting: so digital artists can go back to traditional methods. Like a game where you scan in crayon textures. It's a unique look. More handcrafted.
Maybe AI will bring about a sea change in how we express these traditional works in the digital context.
It's true we miss a huge chunk of art by just considering pixels. Museums are dedicated to the idea that art and expression are intertwined, and that context--both past and present--brings a unique experience to the observer.
Curators certainly aren't generating those longform descriptions next to art pieces. Someone had to think deeply, analyze, and type that out.
except that I can take an AI generated image and burn it into wood or acid etch on metal with a CNC machine, or 3d print it.
Then youd be an artist at that point.
AI will make many, but not all, artists redundant.
As usual, the lower skill/value work will be automatized sooner. Platforms like fiverr will probably suffer a lot, but so will many "for hire" artists. High value work will remain to be exclusive for humans.
Many high value artists (programmers ...) started as low value and worked themselves up. But if the low value segment is eaten up by AI then the path will be destroyed as well. I wonder if this will make classical education more valuable (after graduation you're starting as high value producer) and further increase class division.
I don't think education has anything to do with this. AI tools are pretty easily accessible to anyone and you don't need a "classical education" to wield them.
If anything, I think the necessity of education is going to be weakened further and more quickly in the next decade.
I think they point they were making was that a job as a lower-skill artist was a potential on-ramp to a job as higher-skill artist.
It isn’t obvious at this point what the skill ceiling is, on using AI tools, since they’ve only been around for a couple months.
What a nice sounding dystopian future you've described there.
5 replies →
> I wonder if this will make classical education more valuable (after graduation you're starting as high value producer)
I haven't noticed a correlation between formal education and the people that I would describe as "high value producers". That could be a bias because the median person without a degree doesn't get hired at all, but if we're removing mid-tier and below devs from the work pool, that's going to eliminate most of the people I know with CS or SE degrees.
> I haven't noticed a correlation between formal education and the people that I would describe as "high value producers".
I agree, but I would argue that people without formal education have to usually work their way up.
Basically, if I'm a high school drop out, then I'm very unlikely to land a job at FAANG right away. Quite possibly I will have to start in some low value job to get some credibility. But if AI eats most of these jobs, then it also removes the possibility for the high school drop outs to move up.
Only society can make artists redundant. If people want to get their "art" by digitally rehashing what real artists did in the past then we don't need artists.
How exactly things will play out is not clear: photography did not eliminate painting. Algorithms will not eliminate more "manual" creative work. Some new genres might emerge.
Ultimately what is a more important problem is that the commercialization of artistic production was always challenging. When you can create infinite replicas with semi-random variations it only makes the problem worse.
Exactly. Art as an economic industry may die out. As a career, it may be reduced to a much smaller scope. But we'll always be able to express ourselves artistically - maybe even more so if AI frees up more time in which to do so.
...or enjoy making art with other people to use in our commercial endeavors ?
Like, it might actually be fun to work with creatives to make art. It's one of the areas of my work I actually really enjoy. I can't imagine prompting DALL-E to be as fun and enjoyable, nor would it be "creative".
1 reply →
Commercialization of anything and everything is the real capital P Problem that needs solving, in the grand scheme of things.
While people are (to varying degrees) generous, its hard to build everything on generosity and recirocity. On the other hand once you start bean counting and transactionaling everything and in particular not even accept different types of beans, out entire existence becomes the one-dimensional, money driven disaster it has become.
Many artists are really scared by what these apparently uncontrollable interests are unleashing. Who can blame them.
1 reply →
I wouldn't say thanks to information theory, but rather on human tendency to be bored with repeated styles. It doesn't matter how good the tech is, but if it comes out like Michael Bay movies or anything else that we learn to recognize, people are going to want 'more original' art. The other difference is that the point of art is to express and evoke emotions, it's an open question whether that can be done effectively without having feelings during the process.
This is a good take and under appreciated generally. Already theres an identifiable generic flavour to much AI art. Just knowing that it’s AI generated makes it feel somewhat lifeless. Not because of the end product is lacking , but because of the awareness of the production method. it’s less engaging if you know a machine spat it out, especially once you recognise and become bored by the stylistic markers that indicate machine generation. That will play into the spending calculation of, for example, advertising agencies, particularly high end brand work.
I’ve seen some screenshots of ai ads already. They are usually somewhat horrying, like the person will be smiling, but you start to notice their mouths are open too large and they have too many teeth and lack eyelids, and the people in the background are even more distorted and nightmareish.
1 reply →
Bingo - solving the problem of making "good" AI art seems close to solving the problem of making AI human.
I wouldn't make that strong a statement anymore. The level of play by AlphaGo/Zero seems to demonstrate creativity and understanding that we would only attribute to humans. That is to say I can't say for certain that emotion couldn't be faked, and we don't know that human emotion is super-special, only that it's complex.
2 replies →
"AI won’t make artists redundant"
Of course it will make them redundant. All it takes is one "creative" in place of several graphic designers. What you write are just theories. Do you know what the design process looks like? How many "human problems" are there along the way?
Only the ARTISTS will remain. Artists whose work you will want to buy. Such a fancy for the rich.
I hire artists to turn my terrible sketches into decent looking stickers, generally in a fun, cartoony style.
For the last two ideas I tried midjouney and the best I could get from it was an additional reference to send to an artist. I watched a few YouTube tutorials that offered some interesting techniques for getting better and better images from midjouney, but to my eyes those midjouney experts seem awfully flexible about what the final output will be. They’re dancing with the AI but not fully wielding it.
Perhaps next time I’ll hire both an artist and a midjouney expert and compare results that way.
If it's just a character, you want Stable Diffusion. If it's a "scene" then you might want Midjourney.
Look into ControlNet. Midjourney is not very customizable. Checkout reddit.com/r/StableDiffusion for examples of controlnet. You can turn your sketches into art with it.
That is a very good point: having tight control on style and composition is significantly harder than getting a good-looking image with those techniques (it is doable but it takes practice to learn how to engineer prompt accordingly)
Wish I could filter this conversation to those who are actually artists by profession. A little hard to hear a whole bunch of non-artists telling them what they should do or feel about this.
This is a strange line of thought. I mean, yes we'll get a whole bunch of bullcrap posts from people not involved, but effectively you're instead reducing the conversation to "Lets give the artist their 15 minutes of neo-luddism".
Artists are not the first group to fall under the hammer of technology and automation, and they will not be the last. I'm sure most blacksmiths thought themselves artists that worked hard to master a craft after years of work, and now a press stamps out that same work in seconds. This is hitting people hard now because things we think of as distinctly human are now being accomplished by machines (though was this not true in the past?).
There are discussions for artists here, but there are plenty more for all of society. Jobs will get replaced and change in form at an ever increasing rate due to technology if trends keep up. Will the rate of technology change job requirements faster than humans can retrain? If it does what are nations and societies going to do about this. Much like the AI safety issues, we need to answer these large scale issues now before artists and programmers are stabbing each other in the streets for breadcrumbs while multitrillionares that own the technology live like gods.
That's interesting. HN has a software bent on startups and such. As if programming is one of The Known Ways to create something. Fundamentally, our instructions are executed by a machine to some effect, one of which could be profit.
I wonder if artists are having their "software moment": a decade from now (or sooner) they will say, "Adobe Photoshop is the assembly language of digital art; use these AI frameworks/prompts/GPU cloud mix to do X and Y."
From that perspective, it's just pragmatic. No one writes assembly unless it's either for passion or practical.
I think that “AI” (there is not even a drop of an actual AI in the current technology) is overhyped and barely can handle any of jobs people wanted it to do. It’s maybe good for sketching, be that code, text or images, but not for final product, and won’t be any time soon. As it was with all “AI” products in the past, it’s over advertised by ML bros and grifters.
Half an hour trying to force one of the models to spit a manga-style picture of a character... with correct number of arms and legs. No successes so far. It's frankly disgusting most of the time, prob due to uncanny valley-like effect - it's kinda close, but not quite there. Not to mention, I'm sure I did not put "amputee" in the prompt, but I got a few pictures that should be honestly tagged with "guro"... and I wasn't even trying for NSFW!
Same experience with code: Github Copilot is 90% right 100% of the time. Its suggestions for comments and docstrings are so bad I would never accept them in a code review, basically your old "a++ # adds 1 to a"; the generated code is always wrong, and the bigger the chunk of code generated the more wrong it is. It's kind of OK as a replacement for Ctrl+C/Ctrl+V and for generating boilerplate... which shouldn't be there in good code anyway.
I suspect that for every impressive output we see there are tens of thousands of trash outputs someone had to wade through. Further, the less popular your prompt and the smaller the representation of what you want in the model, the harder it is to get something even remotely resembling what you want. Though, I'm not a "prompt engineer" (WTF is that...), so maybe I'm just using it wrong...
You guys need to understand, that the skill ceiling for AI art is very high. Its not some iphone-selfie tier technology. Go to Civitai.com and huggingface, to start comprehending how many AI models there are, and how powerful the latest models are. It takes at least two weeks of intense usage to get the hang of the basics/parameters and produce good output. AI art in the end will probably be dominated by professionals because of the increasing skill requirements.
1 reply →
Pray tell, what are you going to do the moment it is not overhyped?
Sitting around on our asses going "We don't have AGI/ASI yet, nothing to worry about" is kinda like saying 'nuclear bombs are nothing to worry about' in 1940 while watching your neighbor create an every larger pile of uranium. To think we're not going to accomplish this eventually is foolishness, and when we've already accomplished it, it's going to be very difficult to setup a legal framework that reigns in corporate interests around it.
What are you even going to do? We have no agency to do anything with the march of society and technology. Business interests rule and we are fed information that aligns our interests with those as much as possible. I mean climate change is a much more real evil than startrekian fantasies about ai, and yet no one here is doing anything or marching in the street or blowing up oil refineries. There should be ten threads a day on the environment here if people gave a crap like they should. That should give you a preview of the level of political engagement of the current residents of the earth.
I’m gonna be doing whatever I’m gonna be doing. Job market changes all the time, and people adapt. I feel pretty secured (at least for some time) for my field, which is a lot of R&N in wireless networks/radio and embedded systems. My point is that current “AI” is just very well learned models, but it can’t really do intelligent work on its knowledge, can’t innovate, can’t infer, doesn’t have an actual thought process.
Founder of dreamphilic.com here. Based on our users I am seeing some trends: 1. Great artists are generating great prompts and thus great images because at the end it all comes down to your imagination. 2. I expected most users to be young population but what I see is they are mostly people of age 35 and above. 3. Creative people are enjoying it. People who are not good at using artistic tools or don't have access to those expensive tools but have good imagination can also see their imaginations come to life, thanks to AI.
I want to make a cartoon. Who can produce the imagery faster? Either spend $ commissioning an artist or spend money entering text in to a computer.
"A teddy bear walking though a park". If I don't choose the artist, the artist is out of the job and the AI wins.
You could say someone needs to train the AI with an artwork dataset but once that's initially done, its over, considering that studios are the main supplier to artists income.
Does this not cause redundancies?
Be realistic and practical here, in what situation are you going to start typing into a computer "A teddy bear walking in a park" and then start to do anything with the images it generates in a case where you would've entrusted and artist or designer to make something for you ?
What about when you need different scales, different colors, different this and that, are you going to sit there and keep prompting for this ?
I've worked with people who are really good with Illustrator, they can change and produce images for different purposes in a ridiculously short amount of time already. The ting is, even if I could recruit AI to do it, I probably wouldn't and just want someone else to take care of it. To make favicons and screw around with thumbnails and vector images and the list goes on.
There's also the fact that, some people have an eye for style and application of it. Even when I've tried to use DALL-E, which I never have successfully because it always generates something weird enough to be unusable, I've not been confident that my own personal selection is good enough, I'm not a designer, I don't have an eye for art or color, or style etc. It's not my profession to know what's nice. I don't want that responsibility, so I outsource it.
The problem arises if a bunch of studios start adopting AI assisted workflows to increase productivity and the studios that are holdouts get left behind because it would no longer be economically viable to continue animating without AI. It could be similar to how traditional cel animation has declined in use. If AI assisted output clears the bar for consumers, then it's probably good enough for managers.
People using AI professionally probably won't put the raw output of DALL-E straight into the finished work either, I've seen artists that use Stable Diffusion to generate a base image and then do heavy amounts of editing with Photoshop or similar.
1 reply →
AI won't make artists redundant because it does not experience human sensations. The only thing an AI can do is read the flow of data we make available to it, which is text, audio and video.
An AI will never be able to represent what it means to be subject to police brutality, because it cannot feel being beaten and humiliated by a cop.
An AI will never be able to paint an anti-war poster from the perspective of someone who has felt the ground shake due to shelling of their hometown, because an AI has not lived there for 30 years and does not know the feeling.
The only way AI will make artists redundant would be if it became completely human, which is contradictory since at that point it would be a real artist
In case you don't know, AI art is not an AI pumping out art by itself. A human prompter has to prompt the AI. The prompter can perfectly understand 'police brutality'/'shelled hometown', as they are human, just like the artist.
Artists feel threatened, because they invested 90% of their time in drawing skills, which went from unthinkable to automate, to AI surpassing 90% of artists in 8 months.
> they invested 90% of their time in drawing skills, which went from unthinkable to automate,
Except that the AI doesn't automate drawing, it automates the creation of an image of a drawing.
I have a friend who is an amateur water colour painter. I have no doubt that an AI/ML application can produce pictures that look very much like pictures of his pictures. But at the moment at least it is utterly unable to create the actual artwork.
All of that is true, but as soon as you model subjectivity and embodiment with any kind of credibility it's game over.
The current generation of tools doesn't do that. It's essentially a very sophisticated parrot making speech-like noises it doesn't understand.
But a couple of generations from now, I think it's going to be much less straightforward.
Also, the thing about genius is that it's subjective. In the arts it's more or less synonymous with mastery of a medium with impactful novel insight.
You could argue that AI systems are well on their way to mastering visual media. It's not quite true, but that's because the people training these systems are not artists, and so far they're selecting work that looks a bit Social Media and Game-Ish rather than Art Museum and Contemporary Show-ish.
I don't see any reason in principle why that couldn't be fixed with better training.
So what about impactful novel insight? The point here is there's a kind of cultural and perceptual feedback process which selects certain works out of semi-random cultural noise. There are always a lot of artists making a lot of work, and most of it is not that interesting. Selecting interesting work doesn't require intent, it just needs a feedback loop.
So if you create a situation in which a community rates the art and selects certain works/algorithms/training sets over others, I suspect the impactful and novel insight will happen automatically.
Sentience or subjectivity are not required. In fact you'll get an automated version of what happens already, where different communities with different levels of education and sophistication select different kinds of work for their own reasons, and some are considered "works of genius" for reasons that may be as political and cultural as artistic.
You are discounting fiction then. Lots of art describe or depict things the artist never personally experienced.
It's not that clear-cut. Humans can empathize, to the point where observing someone getting hurt flares similar neural response to the one in the hurt person. Sufficiently vivid imagination in conjunction with a lot of research can give you an experience "close enough" to the real deal to be able to write about it. I've witnessed an author falling into a deep, clinical depression solely due to the subject they decided to tackle. Not every person is capable of such a deep dive into an experience they nominally don't have, but authors and artists tend to be able to do this. In such a setting, "personal experience" is a fluid term, not necessarily synonymous with "he was there at the time personally" or "it happened to him personally".
When you wake up from a nightmare, you're covered in cold sweat, you have trouble breathing normally, your hands are shaking - did you "personally experience" what you've dreamed of? Authors and artists are in a business of dreaming like that while awake, and sharing those dreams with others.
The AI will surely get there at some point, as others noted, once you model embodiment and imagination it's game over. But it's not as close as others seem to think, in my opinion.
You’re being downvoted for contradicting the AI hype machine, but you are completely correct. Humans have been creating art for at least tens of thousands of years, and none of the reasons we create art will ever be fulfilled by AI.
You are reading too many coffee table books.
You are reading too much Twitter
3 replies →
AI just needs to be better at creating art from instructions than a hired artist is.
And we already know great art can be created following a prompt - hired artists interpret their client's instructions all the time (concept art, game art, movie art, etc.).
hired artists interpret their client's instructions all the time (concept art, game art, movie art, etc.).
I cannot disagree with this comment any more, I hire creative people to help me to create things which isn't in my skillset. I don't hire them to tell them what I want entirely. Sometimes I'd at least describe what I want, but often they will say, well that's good but how about...
That is a huge part of the value add for me personally. If I knew what I wanted, half the battle would be one, AI or not.
I don't think we disagree. Artists help you turn a relatively short and vague description into a finished artwork. The clients usually already provide very little information in comparison to the number of decisions needed to complete the work (e.g. choosing composition, theme, style, colors, lighting, etc.).
Working from vague prompts is exactly what the current AIs do. Follow up questions and ability to do precise edits are likely next steps in the evolution.
I think artists will just have to leverage the tech on a larger scale to be effective. No matter how it's created, the best art will still have value..
It may not be economical to, for instance, have an entire studio of dozens or hundreds of artists produce an animated film with an anticipatory budget.. but I suspect it will still be quite profitable for a handful of artists and some AI to produce that same film in a shorter time period.
Or it may be profitable for a 13 year old kid to produce that film in his spare time. In that way, I see it as a sort of artistic renaissance where new voices are found and the pace of artistic acceleration is unprecedented.
I wait when game from like Dark Souls series is released and all concept art is generated by AI, to me it seems such game will have no style consistency, it will look like game that is is put together from market place props (eg. cheap garbage).
Because AI can generate very good looking art works, I don't think it is all there in making compelling, consistent, and visually pleasing art, simply because there are feed back loops, between multiple people, from concept art to end product.
- 3 years later -
The whole game is generated.
It brings on a renaissance of game designers, because now ideas can be iterated and polished instead of dumping thousands of hours into prototyping.
AI can create the skeleton (for now), but we need new AI models for the meta-analysis (game balance, replayability, novel mechanics).
And AI game reviewers to curate that, because we won't have time to play everything.
What's sad about AI is peoples reaction to it. The defensiveness from artists misses the point: Technology should better our lives and free us to pursue our passions. The real problem is under our current economic model, it has the opposite effect. Instead of getting defensive, we should be rethinking the role of our economy. Sadly (and cynically) I suspect we won't see change until AI comes for the CEO's and politicians.
>an open-source Stable Diffusion, which is a basis for multiple projects, including commercial Midjourney.
source? The model behind Midjourney looks quite different from Stable Diffusion, to me it looks like a model conditioned to image embeddings (like Karlo or Dalle 2) instead of text embeddings, also I don't see any mention of the Creative ML OpenRAIL-M license on their website, when they experimented with SD in August they added it but it has not been there for months, To me it is clear that MJ has trained a new diffusion model from scratch, it is also noticeable by the difference in performance, even for a model trained in AI image recognition it struggles with images generated by Nijijourney (MJ but finetuned for anime) while it has an easy time with anime images generated by SD and its finetuned models [1], which means that the distribution of images is very different.
[1]: https://huggingface.co/spaces/saltacc/anime-ai-detect
I think so too. AI might affect applications like Photoshop if the result of the provided output is good enough. Stable Diffusion is going into this direction. While ChatGPT is nice, real authors and journalists create value by bringing in a nuanced and subjective insights based on their research/expertise/ideology.
Personally, I highly value the vanilla perspective of chatGPT's explanations compared to picking through some authors bias to try to find truth for myself. What you are calling value is nice for entertainment purposes or perhaps some heuristic understanding and essentialization... But it is usually at expense of nuance and truth.
It doesn't matter whether diffusion programs or even more rarefied approaches to solution spaces can capture the essence of human-produced art.
Markets are fundamentally about perceived value instead of actual value. Algorithmically generated creative works do not have to compete in the field of actual value; they has to compete in the field of perceived value. Let's be real, almost all customers of art - whether they pay before or after the production - are pretty bad at estimating actual value.
More philosophically, it's not possible to empower humans above human superorganisms. You make a successful walled garden or you fail to reproduce your ideals into the future.
That is a bunch of minced words to explain a very simple concept:
People judge results, not effort.
You don't buy a product because the producer put in <X> effort, you buy it because it's good (FSVO good).
> People judge results, not effort.
That may be true for graphic art for movies, video games, websites, etc, but not for fine art. In fine art, the resulting artwork is judged much less by the result than what went into it. The artworld already is deluged by art, so the narrative around the art and the artist is the most important differentiating element.
Food Safety and Veterinary Office? Fire Service Vehicle Operator?
1 reply →
What is "actual value"?
In fine art? Social signalling. AI generated art is like artificial diamonds.
2 replies →
I have a hard time being worried until AI actually becomes comparable to human intelligence (and then art will probably be a lesser concern amongst others). Why? Because art is about the generation of things that people find appealing, which you can automate, but it's as much about am emotional conversation between people (using a "language" words can't replicate), interest in artists as people, and doing novel things. My suspicion is that whatever great thing people can do with AI, in the ways it exists now, people will get bored with it very, very quickly and move on to something they feel is more soulful.
The valuable art will simply become more performative. We still pay a ton for live music even though we can hear the same music for dirt cheap.
We pay a ton for theatre when we can just watch a movie or recorded play.
We have adapted to cheap art many times before.
What I've seen from all the AI lately is a lot of vague uncanny valley images that somehow don't fit. They're too perfect and even when they try to be imperfect it's too intentional.
AI won’t make artists redundant because you don’t pay an artist for a picture in the first place, you pay them for their creativity, their ideas, which may eventually result in a picture. You know, their artistry. Too many artists devalue themselves and their livelihood by defining their work as entirely mechanical, entirely defined by their material product. Art is more than just produce. You’d think artists would be the first to say that.
The value in artist is their skill in realizing their creativity. I can be creative too but but sure as hell I can't paint it.
> Will there be room for human artists?
When the details of the machine's output will have a justifiable foundation, then the machine will be an artist.
Before that, you employ «human artists» because they have those missing modules.
The need is in those missing modules. The agent who has them can "do everything" or cooperate with other specialized agents - like in normal work.
Imagine a world where AI art tools are commonly used.
Ignore the messy legal and moral issues; just imagine, for a moment. What's changed?
These tools effectively enable artists to:
- generate a variety of concepts for some desired output
- render out the detail, lighting and shading that has traditionally been done by hand
- convert existing images and concepts into artistic 'styles'
You want a set of icons? You want a custom font?
The effort is reduced from weeks or work to days, if that.
People will have to learn to use them; just like any other tool.
So really, what actually changes?
Two things, specifically spring to mind:
- The existing skills that people have to do these things become obsolete.
- The number of people needed to do the current job that artists are employed to do is drastically reduced.
That's what's going to happen, and it's happening already. You need people to paint tween frames for your animated video? Well, 2 guys in a basement can do it now instead of a team of 10.
Yeah, you still need the key frames... but the mechanical shading and drawing that employs of lot of people, specifically in the animation industry, is going to be ERADICATED.
In corporate teams, will your 'design team' of 5 people be cut to 1? Probably not.
...but, those industries don't employ the majority of creatives. The majority of artists do not draw concept art. They do mechanical technical processes in VFX and animation.
There will be new companies, and new roles.
...but, a lot of people will find the mechanical work that earns them their daily living wage will replaced by a much more efficient automated process. Those people may, perhaps, be able to re-train and get new creative roles doing other things.
...
What the 'anti-AI' movement has won is a short reprieve until the 'ethical' and 'watermarked' corporate AI art generators roll out, and get accepted as industry standard. Maybe that's a good thing? I dunno.
I think it's fair to sat that the cat is out of the box now.
In the future, there will not be industrial scale human drawn 2d art. That industry will no longer exist. Hopefully the people currently in that industry can find other creative roles (or retire) by the time that happens.
When photography came out, it had a similar impact on hand drawn photo-realistic paintings; and I think you can see thing, historically. It's not that no one does it any more. It's that it has become a niche job, that is technically inferior to just taking a photo.
That's where this is headed for 2D artists.
>What the 'anti-AI' movement has won is a short reprieve until the 'ethical' and 'watermarked' corporate AI art generators roll out,
They've managed to get some NSFW models shut down, but have they achieved anything else?
MSFT looks like they want to go full steam ahead with this, and they have pretty good lawyers who will not be frightened by a few frivolous lawsuits.
I think the case of https://arstechnica.com/information-technology/2023/02/netfl... shows that the movement has had a tangible delaying effect of becoming industry standard.
I don't believe they'll be successful in stopping things from happening... but I think it's pretty undeniable that it's caused things to slow down in wide scale corporate adoption.
1 reply →
Great artist will steal from AI and leave the 95% of artists redundant.
it sure as hell will displace artists and wreck art market. people are already choosing to pay couple bucks for an app to get some "art" from a random company that repackages stable diffusion, which was built on unlicensed art, instead of paying an actual artist to create something. and when called out on their choice, people make excuses like 'it's just for fun', 'it's cheaper than paying an artist', 'i don't have money for an actual commission' (and yet, somehow they have enough money to pay for an app). apps are currently actively undercutting and pricing out artists, and it is working for the apps, which also get free promo with every customer that gleefully posts their results.
apps can set any price they want on what costs very little to create - there's some compute, storage, etc. - but they can just repackage existing models, train on unlicensed artwork, get all the parts for free, and then price it (and undercut) any way they want. there's a lot of leeway on how aggressively can a service like that price it's "goods" which are basically free to make, and can be made in no time, at enormous scale. there's infinite possibilities for predatory pricing when the base price is approaching zero. actual human artists, who do commission work, which take time to complete, and are worked on one piece at a time, can hardly compete. even an entire market (or a piece of it), full of different artists, hardly can.
That's only a market problem if they are "dumping", i.e. selling under their price of production in order to gain more market share. Like Uber subsidizing their actual product with VC money and gaining more market share. I think you can sustainably sell the images for pennies as long as you run your own hardware and keep a lookout for unnecessary costs/middlemen (AWS, various paid Google products, etc.)
Apparently, undercutting (and other forms of predatory pricing) is not a market problem.
To reiterate, they're pricing out other artists, and undercutting artwork (such as, commission work) prices, not just the price/cost of generated output (which services can also undercut/predatorily price between their own kind, going as low as 'free' - 'but you'll have to pay us extra for priority/compute/features/whatever').
An artist might be offering commissions for, say, $10. Someone on fiver might do something for $5. An app comes in, and offers you a package of 20 "commissions" for $2. See how that undercuts artists? It's frowned upon even within artist/trade circles when artists do that sort of thing (as overt lowering of prices can affect everybody else in the market), let alone when a random non-entity comes into the market to disrupt it (and that is the intent, when such services specifically advertise their 'art-making' ability), and does so while operating on your own unlicensed (stolen) artwork.
2 replies →
How do you vote down a link?
With the flag button. You won’t have access to it til you get 500 upvotes I believe.