bits and bots
On Threads, which I can only take in measured doses, I've been idly following some of the discussions around AI (the actual discussions, not the handwringing and moral posturing about what is "real art," which still make up the bulk of most criticisms.) More the ethics of image scraping. At this point, with over 20 years of making things that could be called art in various mediums(and sometimes probably more on the "design" side of things than great works of genius) I mostly roll my eyes at those who question the authenticity or skills of people who use AI image generation, or embed it inside other creative work. Whether they are "real artists" or "posers" playing with a new toy. Granted some people are just having fun, as they should be able to in a world as messy and chaotic as this one. Others have more specific outcomes they are looking for. I've often, however, gotten the most interesting results entirely by accident.
Sometimes, I am just running prompts and tweaking them to get exactly what I want, which sometimes turns out awesome maybe 10 percent of the time, 90 percent not so much. I wouldn't say its hard, which I feel like some people distinguishes "real art" from "fake." It feels more akin to creating a website or using found graphics for book covers. Or maybe even more like curation, running and changing the prompts and then choosing what I want to use. It takes a little bit of knowledge of terminologies like camera types and processing, but even here I only have a little bit. More it takes having some sort vision and direction for what you want to produce. Or having that elusive thing, a "good eye."
Again, much more "design" than "ART."
The handwringers will tell you that AI will be soon eliminate all artists, which doesn't seem possible. I'm not sure anyone who fears being replaced by some code is actually creating from a place that is all that genuine in the first place. You still have creativity and vision. Nothing can really take that from you. Still want to express what you were put here to express. The discussions feel a lot like painters worrying that photography would eliminate them. Or that film would eliminate photography. Of course, it all comes from a scarcity mindset rather than an abundance one. While image generators are far more sophisticated than other LLM's (the results of bots writing poems are about what you would expect from high schoolers who've not read much poetry beyond the 19th century.) The one field where the fear may be genuine and concerning would be non-creative types of writing can be bolstered by AI's tendency to just get things wrong, make shit up, and just in general sound weird (or bad at best.) Not that they are replacing "real writers" who will always be necessary for any sense of voice or vision, but more the rampant spread of misinformation it may create in their absence. Still, I don't blame the AI, I blame how people may misuse it for deception.
I understand a bit more the artists who are angry that venues like Google and Meta are using their work to train their models, but really, I've long suspected that everything anyone ever posts on the internet, while it may remain yours by basic copyright, becomes part of the internet consciousness and therefore moves a bit beyond your control. It's kind of already in the terms when you join many sites. Someone I saw today that this actually a good thing for culture in general, that the internet collective should be vast and varied and a stew of everything happening. That its a little cool to be part of that collective (since I too have images everywhere online that are part of that dataset.) No one should steal your work and pass it off as their own, obviously, but everything is influence and very little is new under the sun. Sometimes I will glance at the most ardent naysayers and they are totally violating trademarks by drawing like Pokeman fan art or some other big IP, so I kind of take their arguments with a grain of salt when they talk about "stealing" and "theft."
I think a lot of discussions lack a certain nuance and understanding how LLM's work to create images. It's not like collage, where I quite regularly work with appropriated graphics and images. I try to stick to public domain and stock imagery to not step on toes and honor copyright, but everything else is fair game for collage fodder. The only exception often being physical collages, whose materials I sometimes lift from magazines and books willy nilly without attribution, as generations of collage artists have done before me. And really, most of my sources and materials are vintage anyway and not even related to contemporary artists.
Generators like Midjourney, of course, produce things based on the language of the prompt and the AI's understanding of what something looks like. So if I issue a prompt for a woman in a white dress, the algorithm of the AI creates an image based on its understanding of what this is supposed to look like. It's not pulling from any one or even 100 artists, but a collective understanding of what constitutes "woman" and "white dress." And its sometimes just wrong or weird about images, especially when you start adding in stylistic prompts and types of photo. The woman and the dress don't exist elsewhere, but they exist now based on those parameters and understandings. It's not another artists work, but they did potentially provide the dataset that helped create it. Many feel they should be compensated as such, and this would be awesome, but I don't really think that will ever happen. Because capitalism and all. Most actual "theft" is intentional, done by unscrupulous rival artists and corporate greediness, not the result of 1000 monkeys typing Shakespeare
 |
| totally real photo of the Jersey Devil... |
I may be alone in thinking this all makes AI a great tool for world building and story generation, where you have greater control and diversity in what you can use without violating copyrights for mimicking other artists. And, since most of my work exists a bit in unreality, I actually appreciate what some people criticize as the uncanny effect of most AI art. I want it to look artificial, and it does most of the time. But then so does a lot of painting and other mediums. Some things look very real, so it's actually things like deep fakes and public manipulation we should be worried about rather than some AI artists funny "hedgehog in space" image. Of course, I am also a little artistically mischievous, so immediately did things like make faux cabinet cards of cryptids..lol..One of my favorite art exhibits we ever did at the library was about hoaxes, including an exhibit that staged a full alien autopsy down to newspaper clippings, so I think there is room in art for it as long as you're not trying to, ya know, overthrow democracy or slander people.
There are also discussion of copyright when it comes to the generated images, having been created by a machine, do the generators have a claim on them. Probably not. I would not consider my AI experiments to be "mine" anymore than using a clip of a public domain video in a videopoem is mine. But I can make it into something else. Add context and purpose and create world or a story. While some of my AI generations are just for fun and inspo (see image above), some may eventually, like any stock image, become paired with text and poems (The Alice project is something that uses AI generated images that mostly, outside of some editing things and removing extra fingers, are as the bots intended.) Other pieces, like the Poe tarot cards, use a mix of found graphics and AI elements arranged digitally. Today for a book cover design, I simply needed a basic unicorn illustration with specific colors in its mane. I initially did it crudely with found graphics and it would have worked fine. But then put it in the hands of the bots and loved what they generated, but then inserted it in the design that I composed originally. It's much less creation, more collaboration with a bodiless artist who always agrees with me and with much better Photoshop and rendering skills. I can live with that.
Comments