AI generated art

[Laura R Hepworth]

@THX1138 Glad to have been of help . As an artist, this has been a topic of particular interest to me over the last year and it's been a difficult slog to understand it. I still don't understand it fully, but I've learned enough.

 

[JunkMonkey]

Porn is going to be the most extreme area for this--because soon you can literally do anyone in a porn scenario that will be about as realistic as is possible and that will create a lot of problems legally and also expose the public to things that are extreme and probably better off not seen. If you try to limit the visual information available to the AI system, I don't know--I just think it is going to be scenarios like: Colossus, do the Last Supper as an orgy with current members of Congress naked and bloated, in place of the apostles. etc.

Rule 34 - I won't post the link, but it's out there.

 

[Stephen Palmer]

"SFF Chronicles" put into Midjourney. I definitely think Chrons should be re-named Shal Fnacles!

 

[Toby Frost]

I've been waiting for Shahil Fhalinces' new book for ages!

 

[Moonbat]

I've been having great fun with Midjourney recently (see my new avatar)
Also the kids wanted battle guinea pigs.

And I have a story I tell the kids on the way to school about mist witches.

 

[Astro Pen]

"SFF Chronicles" put into Midjourney. I definitely think Chrons should be re-named Shal Fnacles!

View attachment 98740

Bottom left image looks very Chris Foss

 

[Stephen Palmer]

Bottom left image looks very Chris Foss

Yes, it does. Perhaps a lexical connection in the prompt between SFF and Chris Foss' name.

 

[KGeo777]

I saw some really cool images--one was a what if Alexander had conquered the world.

What would it look like.
Also what if the Egyptian or Roman Empire had never disappeared--what would modern police look like?

 

[Victoria Silverwolf]

I fool around with sites that allow me to play with stuff like this for free and without signing up. I'm not serious about this, and I certainly have no intention of making any commercial use of the images, but it's fun. Just now I generated several images based on a description of my imaginary self. They came out very nicely.

Here's one example.

 

[KGeo777]

Imagine if this technology was around in the 1980s. It would be like magic.
If they allowed these AI things to have access to Getty Images---it would be like a holodeck of imagery power.

I saw a demo of fake people made entirely by computer--the resolution was standard definition but the faces were totally convincing.

But with Deep Fakes and this--it's only a matter of time before someone develops a computer program that will create new episodes of 1960s Star Trek. Guaranteed it is going to happen.

One of those computer geeks will be a big Star Trek fan guaranteed.

 

[Astro Pen]

And AI 'deep fakes' video pretty well these days. This is me as Bond.

 

[Victoria Silverwolf]

Trying to come up with images of the continuing sword-and-sorcery characters Kha and Volen I sometimes use in the writing challenges. These aren't bad.

 

[Victoria Silverwolf]

Still fooling around with this, I asked one of these sites to come up with "A woman named 'Victoria Silverwolf.'" I am pleased with the result.

 

[Laura R Hepworth]

If they allowed these AI things to have access to Getty Images---it would be like a holodeck of imagery power.

They do use images from Getty (among other places) just not with permission. That's why Getty is suing Stability Diffusion. They scraped over 12million images from Getty without permission.

 

[JunkMonkey]

I've been thinking about this and I've come to realise that the thing that really irks me about the (general) discussion of AI generated art is the inclusion of the word 'art', or possibly the word 'generated'. Or both.

'Art' first.

This is not 'art'. It's AI generated imagery. At best it is illustration (in the way that a pie chart or a map would be). These things (images) are generated by writing text and feeding prompts into a machine which converts that text into images by pulling out images associated by humans with those (language specific) words and mushing them up via some complex algorithms. None of them come about as a response by an AI to the real world or - more importantly, other pre-existing art - not in a way that humans make art. No AI has yet, as far as I know, needed to make any of these images. It makes them because it is told to and then told what to do.

'Generated'. At best AI could be described as a tool to be used by an artist. No one talks about 'paintbrush generated art', or 'pencil generated art', for very good reason - there's no such thing. Artists make art. What they make art with varies. The pencil is as creative as any other inert object. So is an AI. Three year old toddlers produce art. AI's don't. They don't produce art any more than a pneumatic drill will drill interesting holes in the road from its own volition, or adjustable spanners take it upon themselves to tighten nuts in new and provocative ways.

Ok, yes I will say some of the images produced are beautiful, some are intriguing, some are baffling, ugly, and frustrating in ways that many pieces of genuine art can be, but then so are rainbows, frogs, rock formations, sunsets, the decay spirals in a cloud chamber and gazillions of other things and phenomena. None of them are art.

The AI itself may be a work of art but I'm not convinced what it generates is.

 

[KGeo777]

That's a good point. It is the human operator instructing the computer to make an image. It isn't doing it on its own.
So to say it is AI-created art--it tends to create the idea in your mind of a computer coming up with the concept and design which is not the case actually.
It is more like a super fancy photo collage machine.

 

[Laura R Hepworth]

I've been thinking about this and I've come to realise that the thing that really irks me about the (general) discussion of AI generated art is the inclusion of the word 'art', or possibly the word 'generated'. Or both.

'Art' first.

This is not 'art'. It's AI generated imagery. At best it is illustration (in the way that a pie chart or a map would be). These things (images) are generated by writing text and feeding prompts into a machine which converts that text into images by pulling out images associated by humans with those (language specific) words and mushing them up via some complex algorithms. None of them come about as a response by an AI to the real world or - more importantly, other pre-existing art - not in a way that humans make art. No AI has yet, as far as I know, needed to make any of these images. It makes them because it is told to and then told what to do.

'Generated'. At best AI could be described as a tool to be used by an artist. No one talks about 'paintbrush generated art', or 'pencil generated art', for very good reason - there's no such thing. Artists make art. What they make art with varies. The pencil is as creative as any other inert object. So is an AI. Three year old toddlers produce art. AI's don't. They don't produce art any more than a pneumatic drill will drill interesting holes in the road from its own volition, or adjustable spanners take it upon themselves to tighten nuts in new and provocative ways.

Ok, yes I will say some of the images produced are beautiful, some are intriguing, some are baffling, ugly, and frustrating in ways that many pieces of genuine art can be, but then so are rainbows, frogs, rock formations, sunsets, the decay spirals in a cloud chamber and gazillions of other things and phenomena. None of them are art.

The AI itself may be a work of art but I'm not convinced what it generates is.

Agree completely. I think, with the current issues resolved, it could be a great source of inspiration for artists or even for authors to get a clearer image of something in their head so they can describe it better (like, for example, a weird creature), but in and of itself it is not truly art. AI can't really think to create, nor does it feel or actually know the fundamentals of art. It can only analyze and approximate images based on the information it has been fed. So, there isn't really 'intelligence' in this artificial intelligence either.

It can be great for mood boarding though. Still, they've got to get the mess surrounding it sorted out before it goes any further.

 

[Laura R Hepworth]

That's a good point. It is the human operator instructing the computer to make an image. It isn't doing it on its own.
So to say it is AI-created art--it tends to create the idea in your mind of a computer coming up with the concept and design which is not the case actually.
It is more like a super fancy photo collage machine.

And, unless they do a whole lot of postprocessing, the prompter is no more an artist than a person ordering a meal from a fast food restaurant is a cook as they didn't make the image either they just gave a list of ingredients to the AI hoping it will come up with something remotely close to what they wanted.

 

[JunkMonkey]

And, unless they do a whole lot of postprocessing, the prompter is no more an artist than a person ordering a meal from a fast food restaurant is a cook as they didn't make the image either they just gave a list of ingredients to the AI hoping it will come up with something remotely close to what they wanted.

YES!

 

[Astro Pen]

I've been thinking about the technological singularity and what steps the road to it will have. Could this be one of them? Part of its toolkit?

An emergent property of technology is the way it now uses humans to improve itself. We are constantly baited into refining it. Already far more addicted to, profiled, and dare I say algorithmically controlled, by it than we think.

Importantly, it is now starting to be creative for us, we don't have to 'create' any more, just improve the technology. It is a subversion.
We think it is 'for us' but it is it's own evolution. It is a natural product, don't be fooled by the shiny surfaces. We have created a carnivore.
Pseudo consciousness and self replication are a bit further down the road.