Then let the people who are so much smarter make their own art and writings, and train their creations on that. Crazy idea, huh?

As Bill Hicks said: play from your fucking heart. It doesn't have to be deep, but it has to be something you express. It doesn't matter, at all, if someone else said the same thing before. If a human says or does something, I potentially care. If a model outputs something I find interesting, I am interesting in my thoughts about it, not in the model, because I know for a fact there is nobody on the other side.

Anything can create novel output. If anything, the opposite is impossible: even if you make an exact copy of something, it will not be in the same place at the same time. So that's no useful criterion for anything, and it's rich to call "sentience" a nebulous concept -- yeah, we can't define it, but we all experience it, and having a name and a complete definition for it would not make it one iota more real than it already is -- while talking about "novel output" as if that means anything interesting or useful whatsoever.

People will inevitably continue to train models as they see fit. Nothing can stop that now. Creativity is a set of algorithms that can be mimicked, learned, and reproduced by a machine. Holding onto a deep-seated preference for humanism is limiting your perspective.

Works should be judged on their own terms, not their origins. And in the near future, the lines between human-created art, collaborative human-AI art, and solely AI-generated art will blur. The point is, soon enough, you will not be able to tell the difference between art created solely by humans, art that is a product of the combined efforts of humans and AI, and art that was made solely by AI.

> Creativity is a set of algorithms

This, following on the heels of mocking the usage of "sentience"? Heh.

But more importantly, I judge things as I see fit, and I like what I like. What is it to you? If I say I only like to be fondled by my partner, and you say soon enough every place will be so crowded that I can't possibly tell if the hand on my butt is that of my partner, of a random stranger, or of a machine, I can only wonder why anyone would find that desirable, if that is supposed to be some sort of threat, and whether you'll enjoy me "reacting as I see fit" to it.

It's not a threat. It's a descriptive claim about what the world is going to be like once this technology matures. In due time, most art will be the result of humans and AI. Unless content has some kind of label stating that AI was involved in its creation, you simply aren't going to notice. Sure, there will be a niche of people who obstinately refuse to use any tech that employs machine learning. And we will view those people the same way we view people who believe that taking a photo of someone steals their soul.

You, like everyone, skipped the question why the people who program the models can't also make the effort to create what they teach their creations, and hand-curate it from stuff given with consent.

Yes, it can be done. People can just take it because they can, and hope to create facts on the ground that way. Kudos to them.

> And we will view those people the same way we view people who believe that taking a photo of someone steals their soul.

For me, art, or what I like about art, is an act of communication -- be that of talking to yourself, trying to communicate with others, screaming pain or joy incoherently into the void. It doesn't have to be words. But I care about something because it's from a human.

> It would not be much of a universe if it wasn't home to the people you love.

-- Stephen Hawking

If people will learn to fake and mass-produce that without oversight, A/B testing for whatever gains they want to get out of it, great, kudos. Seems shitty to me, like something people who aren't creative themselves might do, or who like to fuck with other people en masse, etc. Like destroying something you can't have or make. It's like putting poison into the water supply is possible, but that's not the point.

People can come up with hilarious or interesting prompts, and that is great. And then there is no need to "hide" it, either. But when it comes to the "art" aspect of it, it's still 100% the humans that tune the output (which is an act of judgement, versus calculation) and/or the humans from whom the input was taken with consent, or stolen without.

And last but not least, I take things how I want to, anyway. How I think the author meant it is something I acknowledge and take into consideration, but no more. And I often have several including contradictory readings of something. The beholder always applies the final touch to everything that enters their ken, if you will.

Oh, but it'll be also great for actual AI generated ElsaGate shit (like people used to pretend with EG, to not face how sick that phenomenon was... "oh, it's just AI run amok, tuned for views and toddlers obviously like that stuff, so hey") and for scamming the uneducated and elderly, I'm sure. Just gonna call that right now.

You are veering into territory that is interesting, but outside the scope of the initial topic. All I'm saying is you don't need a machine to be "sentient" or "self aware" to generate novel outputs. Whether or not these models are created in an ethical way is another matter. Even if it was unethical, that doesn't really negate what I'm saying.

> This basically reaffirms my belief that Stephen King a midwit

Good argument against training AI on his books then, I guess.

On the contrary, you absolutely want midwits in your training data set.

He's a midwit because he hasn't solved the problem of what makes humanity humanity?

What does that make you?

No, he's a midwit because his argument implies that he has solved that problem, when in reality, he's just talking around the problem instead of confronting it.

As long as people are reacting to the metaphor, there will be authors who will unscrupulously use and abuse it.

Actually, it has nothing to do with AI. It's a statement by the author that they believe the consciousness is a product of the brain and nothing more. That view reduces humans to machines, the only difference being in the material.

Agree it has nothing to do with AI. Even if we, for the sake of argument, agree that human beings are machines, why should the appearance of human-mimicking reasoning imply sentience above and beyond any other (complex?) computation? Why are LLMs possibly sentient and not, say, pocket calculators?