Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GenAI Art Is the Least Imaginative Use of AI Imaginable

hai.stanford.edu

Post reply on HN
Print view View original

101–110 of 149 posts

Re: GenAI Art Is the Least Imaginative Use of AI Imaginable

#101

Post by TeMPOraL »

voidhorse wrote: post #96

It's because this technology isn't (and never has been) intended for the laborers making the stuff. The ultimate aim of these companies is to address the managerial and c suite people with a product that can replace the labor pool. Honestly, I think the current consumer focus is just an irrational consequence of these companies feeling like they can't leave all that consumer money on the table, but, precisely for the…

You aren't wrong, especially about motivations. The "current consumer focus" has had some weird results, too - particularly in the image/video generation space, it's the consumers, or rather the OSS community, that actually did 90% of the work of turning AI research output into product prototypes - which many a startup now just pick, polish and scale up. Not to dismiss that work, either, but that's mostly the old-school SaaS building work; the hot AI part is already there on Github, packaged into ComfyUI nodes for people to play with.

Full video is a slightly different story because of compute demands, but image generation is basically the infamous Dropbox comment situation - any fancy image generation task you see offered as a service, a moderately tech-savvy person could replicate for personal use with ComfyUI over a weekend. That applies to interactive portraits too - the nodes are but a click away, models run fast enough for local use on an average PC.

I'd like to challenge you on your conclusion, though:

> These tools are purely intended for superficial people who are not interested in creating things, they are only interested in outputs and end results.

But this is exactly how progress is made. Quoting Alfred North Whitehead[0]: "Civilization advances by extending the number of important operations which we can perform without thinking of them." Where for one craftsman, the process is the most important part of their creative work, another craftsman only cares about the output of the former, because it's an input to their creative work. That doesn't make the latter person "superficial", they're just one level up the dependency graph.

--

[0] - https://en.wikipedia.org/wiki/Alfred_North_Whitehead

Re: GenAI Art Is the Least Imaginative Use of AI Imaginable

#102

Post by numpad0 »

ilaksh wrote: post #64

I would really like an objective explanation of why this attitude "that the whole point of art is the process of making it" is anything but pretentiousness in it's purest form. Surely the actual artifact itself has some value? And even if one supposes that a key part of the value is self-exploration (which I disagree with generally because that presupposes an excessive level of self-importance), then being able to it…

  To me, this is a rationalization for disliking AI because it displaces humans to some degree or another. I think that is a fair reaction, but the rationale is nonsense.
People just hates AI generated images. Why can't the framing be this simple?

AI images and users gets filtered out everywhere, closing into an edgy echo chamber groups, and rarely, if ever, for such "it displaces humans" philosophical arguments. AI generated data just evokes negativity, are low quality, and are unsalvageable, and there are little to no prospects of that improving through cooperation because AI-leaning people have ran around setting so many bridges on fire, so its potentials are all but ignored.

If anything, I'd expect the resentment will subside over the next decade, as generation techniques mature and barrier to entry for amateurs exponentially rise. I think we are already seeing signs of such guild moat forming in code generation LLMs.

Re: GenAI Art Is the Least Imaginative Use of AI Imaginable

#104

Post by parpfish »

tommilburn wrote: post #82
ilaksh wrote: post #64

I would really like an objective explanation of why this attitude "that the whole point of art is the process of making it" is anything but pretentiousness in it's purest form. Surely the actual artifact itself has some value? And even if one supposes that a key part of the value is self-exploration (which I disagree with generally because that presupposes an excessive level of self-importance), then being able to it…

> I would really like an objective explanation of why this attitude "that the whole point of art is the process of making it" is anything but pretentiousness in it's purest form. i think that's a problem - i am no philosopher or art historian, but i truly can't think of "objective" truths of any kind when i think about art, the process of making art, or the process of engaging with art. it's nice to make art. i play…

> i think a lot of value is the context of the art itself

I agree. The enjoyment I get out of a good piece of art isn’t just the piece of art itself, it’s that experiencing the art is a way to kind of connect your mind with the human that made it. If part of the art resonates with me, it’s nice to know that there was another person that felt the same way. If the art is a technical marvel, so can think about the genius and skill of the creator.

AI art doesn’t have any of that and I have to conclude that people who talk about enjoying AI art to the same level as human-made art just never bothered to connect with human art on that deeper level

Re: GenAI Art Is the Least Imaginative Use of AI Imaginable

#105

Post by perching_aix »

Ever since AI art came into the scene I've seen more and better artworks than ever before. I understand why people have ethical problems with it, but the results are genuinely really good even when done by beginners.

I honestly think anyone claiming otherwise is just plain lying to themselves at this point. This is to the extreme that sometimes when browsing imageboards I find myself feeling surfeited. It's been a very interesting experience and a bit of a learning moment really.

The common appeals to the ivory towers of high art are missing the point terribly hard. This was never about them, just like typical commercial art isn't about them either. Nobody cares, you just happen to be beating the same drum the masses do right now.

Re: GenAI Art Is the Least Imaginative Use of AI Imaginable

#106

Post by skydhash »

TeMPOraL wrote: post #101
voidhorse wrote: post #96

It's because this technology isn't (and never has been) intended for the laborers making the stuff. The ultimate aim of these companies is to address the managerial and c suite people with a product that can replace the labor pool. Honestly, I think the current consumer focus is just an irrational consequence of these companies feeling like they can't leave all that consumer money on the table, but, precisely for the…

You aren't wrong, especially about motivations. The "current consumer focus" has had some weird results, too - particularly in the image/video generation space, it's the consumers , or rather the OSS community, that actually did 90% of the work of turning AI research output into product prototypes - which many a startup now just pick, polish and scale up. Not to dismiss that work, either, but that's mostly the old-sc…

> Civilization advances by extending the number of important operations which we can perform without thinking of them.

That's for mass production of useful tools. For decades, it's been easy to produce music, write a book, and draw with digital means. The actual truth is that people don't want to invest the time to learn how. Now they can pose as if they've done so, but the result never survives close inspection.

Re: GenAI Art Is the Least Imaginative Use of AI Imaginable

#107

Post by jdietrich »

redeux wrote: post #42

GenAI is not capable of making art. Self expression is not only inherent in art, it’s essential. What AI does is make pretty pictures, but nothing more. It’s not a tool for making art because it removes the means for artistic expression. It’s possible for someone to make art from GenAI outputs, but that requires taking artistic liberties post generation.

>Self expression is not only inherent in art, it’s essential.

That idea is only about 200 years old and strongly Western. The vast majority of cultures saw no difference between "art" and craft, regarding pictures and sculptures as simply decorative objects, of the same basic cultural standing as pottery or textiles. Our understanding of "art" was fundamentally transformed by the development of photography and mass-manufacturing.

Anyone with a glib answer to questions about the cultural meaning of AI-generated art really needs to familiarise themselves with the thinking of William Morris and Yanagi Sōetsu, because this isn't the first time we've had this argument.

Re: GenAI Art Is the Least Imaginative Use of AI Imaginable

#108

Post by mrandish »

I have zero interest in a tool that turns a text description into an image, video or song. However, I would pay serious money for a generative AI tool that worked differently. I make visuals and music because I love the process of realizing and refining my ideas into end results I like. My challenge is the shortcomings in my mechanical drawing skills or music keyboard fingering skills. Things like maintaining correct proportions across an object. Or nailing the triplet note syncopated timing. Things that take months of iterative practice to reliably do.

I need a tool that takes granular inputs during my creation process and collaborates interactively with me to extend or perfect them. Or suggest other options for that one granular element based on my initial attempts. Not finished results, just the next step in the process. And give it to me as an individual stroke, a masked layer or a MIDI track I can then continue to modify and remix manually.

Can I really be the only person who desperately wants this?

Re: GenAI Art Is the Least Imaginative Use of AI Imaginable

#109

Post by skydhash »

crucialfelix wrote: post #52
0xcafefood wrote: post #45

No, Chatbot All The Things is the least imaginative use of "AI." There's even a chatbot to search your Chrome history [1]. Nothing screams solution in search of a problem more than this. [1] https://support.google.com/chrome/a/answer/15248971

Although the chatbot in chrome devtools is really useful. I selected a DOM node and ask "why isn't this visible" and it queried the DOM, css and figured out the height was 0 and explained exactly what lead to it being 0.

The reason for a DOM node not being visible is either the height/width being 0, its position is out of the viewport, it's behind other nodes or its opacity is 0. And that's basically the same for any layout engine that use node tree. The inspector is so easy to use that using a chatbot for the specified reason is a waste of resources.

Re: GenAI Art Is the Least Imaginative Use of AI Imaginable

#110

Post by debugnik »

marssaxman wrote: post #70
redeux wrote: post #42

GenAI is not capable of making art. Self expression is not only inherent in art, it’s essential. What AI does is make pretty pictures, but nothing more. It’s not a tool for making art because it removes the means for artistic expression. It’s possible for someone to make art from GenAI outputs, but that requires taking artistic liberties post generation.

It's not the AI that makes the art, but the person who tells the AI what to do. A painter friend of mine has gotten into generative AI lately, and the sometimes-pretty, sometimes-intimidating pictures she comes up with sure look like an artistic expression to me. She has a vision of something she wants to see, and she uses tools to bring that vision into reality, just like she does with paint; the new tools just give…

> It's not the AI that makes the art, but the person who tells the AI what to do.

Would you still call them an artist if they commissioned another human actually skilled instead of an AI model?

Why can't people accept that their role in this new process simply isn't the one we've always called artist? We don't call computers to the people using a computer, that role is now the machine's. I'd rather label the model itself an artist than its user, as much as it pains me.

Post reply on HN