AI/LLMs AI art bans are going to ruin small 3rd party creators

That’s not my understanding. No future date needed as far as I can tell.

What concrete question is being asked?
It seems there is no argument there. He's arguing a contrafactual with no basis in reality. He's making up a non-existing AI that he thinks that by some process of iterative prompt building, he'll be able to arrive at something that he defines as creative. He's ignoring the fact that, by the act of the AI having to go to the... well, so to speak, of stolen data, it's never going to be creative.

And creating more and more complicated prompts is not going to change that. Fruit of the tainted tree and all that...
 

log in or register to remove this ad

The architect does not own or control the houses built to the design.
The architect’s drawings/designs are protected by copyright though.
The architect will have a legitimate claim if a builder uses the drawings without compensating the architect.

(Usual disclaimers apply…)
In france, even images of the building are subject to the architect's creators' rights. Which is why it's technically unlawful to photograph certain buildings.

[snipped aggro statement]

Which reminds me: Chris Cox, in an interview last week, noted that the AI D&D tools are dead, "the creatives object to them", but also mentions that the artists at hasbro are expected to use AI because it increases their "creativity and productivity"... and that AI characters are "... helping us know if a new design is authentic."
 

In france, even images of the building are subject to the architect's creators' rights. Which is why it's technically unlawful to photograph certain buildings.

[snipped aggro statement]
Interesting. I’ll go to sleep less ignorant than when I woke.

So does that cover all buildings, or those built after a certain date…?
Trade dress? Copyright?
I mean, if you take photographs of certain types of buildings in any country, you’re going to gain some negative attention. 😉
 

What I am talking about is the same as computer programming. I am directing the AI tool through many, many, MANY lines of language to produce an image that is my vision. A computer programmer directs the computer tool through, many, many, MANY lines of language to produce an image that is his vision. The only real difference is that AI is a tool that can understand English, so I can use that language instead of a computer language.
A computer programmer's end goal isn't to "direct the computer tool through... lines of language." A computer programmer's end goal is to create a program (the "lines of language" you reference). Any "directing the computer" a programmer does while creating their program is incidental, and is often done just to check their work (via debugging, etc.). A skilled enough programmer can create a simple program using only a white board, with no computer involved in the process at all.

The programmer might later choose to run the program they created ("direct the computer tool" to run their "lines of language"), but that's independent of the creative process of programming. Additionally, if running their program directs a computer to manipulate a data set provided by other people to produce some output the programmer wanted, the programmer didn't create that output. The program and the people who provided the data set created that output.
 



Not really. Control vs. no control isn't made up by me. I can enter a prompt and just let an AI run wild, having no control. Or I can take an image, even one not supplied by the AI, and then spend hours and hours, changing hundreds or even thousands of details via the AI tool, and have total control.

Calling one "direction" and the other "prompts" is made up by me, because it's important to distinguish between the two. Do you have two more appropriate words so that confusion can be avoided?

Yeah, one is "a prompt" and the second is "a series of prompts." Generally, using a lot of prompts is a bad thing as the LLMs start losing context or just drifting for no good reason. So what you do is have the LLM reduce all your previous prompts down to a single prompt and restart with that prompt.

Just because you "took the long way" to get the same prompt someone else did in one go doesn't make your series of prompts anymore or less valid. Taking more steps is a sign of ignorance with the tools, not a virtue signal of creativity.

My team of developers save that "final" reduced prompt with their last code commit and use the commit diff to validate the reduced prompt. We work with a very simple programming language that use many small, atomic files with very strict formatting and support for external LINTers, which means we aren't as impacted by the non-deterministic issues of LLMs. Being open source software with well documented APIs and tons of public exemplar code from our vendor that the LLMs were likely trained on, its probably legal and moral. Probably.

Of course, we generate 100x more lines of code using deterministic programming frameworks
 

In france, even images of the building are subject to the architect's creators' rights. Which is why it's technically unlawful to photograph certain buildings.
I think it's the same all over the EU. There was a case a few years back where, IIRC, the Wikimedia Foundation had to pay some architects' collection society a bunch of licensing money for publishing pictures of covered buildings. Presumably, this would apply using the same terms as any other copyright issue. There are also some corner cases – IIRC there was a case that determined that since the Eiffel Tower is old enough, there's no need to license anything for pictures of the tower itself. However, at night there are various illumination arrangements and these are under copyright, so you need a license for night-time pictures.

And of course, there are also buildings that are outright illegal to photograph or otherwise create images of, but that's mostly for security reasons.
 

Recent & Upcoming Releases

Remove ads

Top