We really need a AI tag

[Gradine]

As someone who won't even use call AI-generated imagery "art" with scare quotes, I'm totally fine with it being quarantined in a thread where enthusiasts can show off what they generated. I do remember an earlier such thread though mostly being an excuse to post AI-generated cheesecake imagery, which I don't think is appropriate for this forum.

 

[Zaukrie]

I'm still really angry it isn't banned here. It's stealing.

 

[Reynard]

I'm still really angry it isn't banned here. It's stealing.

It isn't stealing -- the training data WAS STOLEN. But how AI generates images (or text) doesn't really work like that. it isn't a collage. It is an algorithm that "creates" via probabilities based on training data and source prompting.

Does that make it "better"? Probably not. But I think it is worth being accurate rather than hyperbolic when discussing these things. Similarly the issue of water usage -- it is far less than anti-AI folks paint it to be, and not actually that different than most industrial processes (all industrial processes use A LOT of water). Unfortunately, because there is so much emotion attached to GenAI, folks opposed to it on principal do not often take the time to learn about it.

 

[Blue]

I'm still really angry it isn't banned here. It's stealing.

It can be stealing. Huge amount of that for training LLMs and AI art.

Or it can be things like Larian training on art assets they own for the new Divinity video game, which is not stealing and something we might discuss here.

If it comes up in relation to appropriate topics I don't think we should censor it.

 

[Zaukrie]

It can be stealing. Huge amount of that for training LLMs and AI art.

Or it can be things like Larian training on art assets they own for the new Divinity video game, which is not stealing and something we might discuss here.

If it comes up in relation to appropriate topics I don't think we should censor it.

Ya, but we aren't posting Larian stuff, are we?

 

[Zaukrie]

It isn't stealing -- the training data WAS STOLEN. But how AI generates images (or text) doesn't really work like that. it isn't a collage. It is an algorithm that "creates" via probabilities based on training data and source prompting.

Does that make it "better"? Probably not. But I think it is worth being accurate rather than hyperbolic when discussing these things. Similarly the issue of water usage -- it is far less than anti-AI folks paint it to be, and not actually that different than most industrial processes (all industrial processes use A LOT of water). Unfortunately, because there is so much emotion attached to GenAI, folks opposed to it on principal do not often take the time to learn about it.

We will clearly never agree on this. That's fine with me.