Random Thread

[Travis B.]

It's an interesting philosophical question. You'd have to think at how humans create? I think you need an ability to feel emotion and enough insight into it to evoke it in others, a will of your own, a consciousness, and not to get all Freudian on you, but most certainly a subconscious.

It's not merely an interesting philosophical question but a serious challenge to humanity and its place in the world. Realistically what happens to human artists and writers when machines can replicate their abilities at a fraction of the cost? The vast majority of them lose their jobs and humanity finds itself forced out of art and literature. Meanwhile there are no better jobs waiting for all those displaced artists because automation has already eliminated work in so many other fields.

The thing, though, as we have repeated to you so many times, is that generative AIs simply can't replicate the abilities of human artists and writers. Consider the term "AI slop" ─ that is the crap that generative AI's churn out, and it is clear that it is obvious to the public that it simply does not match the output of human creators (or otherwise the term would not have been coined and would not have gained currency in the first place). People simply don't want what generative AI's produce, and while generative AI's may seem attractive to short-sighted executives who dream of replacing human workers, it is clear that generative AI's have a reverse Midas touch, and said executives will soon learn that when the content they sell suffers with the public.

Edit: If you are wondering, YouTube right now is pulling generative AI-created content, probably because advertisers threatened to pull out rather than pay money to advertise on channels that people don't actually want to watch.

[WeepingElf]

The thing, though, as we have repeated to you so many times, is that generative AIs simply can't replicate the abilities of human artists and writers. Consider the term "AI slop" ─ that is the crap that generative AI's churn out, and it is clear that it is obvious to the public that it simply does not match the output of human creators (or otherwise the term would not have been coined and would not have gained currency in the first place). People simply don't want what generative AI's produce, and while generative AI's may seem attractive to short-sighted executives who dream of replacing human workers, it is clear that generative AI's have a reverse Midas touch,

AMEN!!! AI will perhaps soon dominate commercial artwork, but most people will insist on human-made art!

[alice]

The thing, though, as we have repeated to you so many times, is that generative AIs simply can't replicate the abilities of human artists and writers. Consider the term "AI slop" ─ that is the crap that generative AI's churn out, and it is clear that it is obvious to the public that it simply does not match the output of human creators (or otherwise the term would not have been coined and would not have gained currency in the first place). People simply don't want what generative AI's produce, and while generative AI's may seem attractive to short-sighted executives who dream of replacing human workers, it is clear that generative AI's have a reverse Midas touch, and said executives will soon learn that when the content they sell suffers with the public.

THIS!!!

Edit: If you are wondering, YouTube right now is pulling generative AI-created content, probably because advertisers threatened to pull out rather than pay money to advertise on channels that people don't actually want to watch.

Not before time! I remember some years ago that, apparently, large amounts of YouTube content was being created by early versions of AI and watched, not by humans, but by bots in order to generate $$$.

[zompist]

So we wouldn't be dealing with a machine at all, but a sentient being. I don't think we're anywhere close to sentient AI. But, anyway, I have no idea what happens next -- all expectations go out the window. What are the ethical implications of forcing a sentient being to write stories for you?

This is the plot of Every Robot Story Ever, and it always ends the same way: the robots rebel, and either war against humanity, or humans have to recognize them as equals. As ever, the techbros read the cautionary SF story "Don't Create the Torment Nexus" and set about building that cool Torment Nexus.

I mean, it seems obvious that we shouldn't build and keep slaves, but these days it's not too clear that techbros disapprove of slavery.

It's a possible scenario, but I'm not completely sold on it. I think that, boring as those may be, there's a lot more creativity involved in a bad CSI episode than we'd expect.

Even at the script level, that's true. As I said in my post, AIs can't even to customer service, though CEOs think they can. Is scriptwriting easier than customer service?

[zompist]

I don't know what exactly these three comments are referencing, but I suspect that there might be people on the ZBB who do. So, what are they talking about? And did whatever they're talking about perhaps help inspire zompist's story The Multipliers? https://zompist.com/multipliers.html

My dude, I'm old. That story was written before the Internet, on a typewriter. Long before Internet memes.

[Raphael]

I don't know what exactly these three comments are referencing, but I suspect that there might be people on the ZBB who do. So, what are they talking about? And did whatever they're talking about perhaps help inspire zompist's story The Multipliers? https://zompist.com/multipliers.html

My dude, I'm old. That story was written before the Internet, on a typewriter. Long before Internet memes.

Thank you!

[malloc]

AMEN!!! AI will perhaps soon dominate commercial artwork, but most people will insist on human-made art!

That statement contradicts itself though. If everyone insists on human-made art, then it hardly makes sense that AI comes to dominate art anyway. Unless you are suggesting that publishers and studios and so forth will simply ignore popular demand and force everyone to accept something they don't really want.

The thing, though, as we have repeated to you so many times, is that generative AIs simply can't replicate the abilities of human artists and writers. Consider the term "AI slop" ─ that is the crap that generative AI's churn out, and it is clear that it is obvious to the public that it simply does not match the output of human creators (or otherwise the term would not have been coined and would not have gained currency in the first place). People simply don't want what generative AI's produce, and while generative AI's may seem attractive to short-sighted executives who dream of replacing human workers, it is clear that generative AI's have a reverse Midas touch, and said executives will soon learn that when the content they sell suffers with the public.

Sure but what happens when AI improves to the point where it can compete with humans in terms of quality? You must understand that the computer industry is spending hundreds of billions every year on improving AI and it seems incredible that such investments will not yield noticeable advances eventually.

[bradrn]

AMEN!!! AI will perhaps soon dominate commercial artwork, but most people will insist on human-made art!

That statement contradicts itself though. If everyone insists on human-made art, then it hardly makes sense that AI comes to dominate art anyway. Unless you are suggesting that publishers and studios and so forth will simply ignore popular demand and force everyone to accept something they don't really want.

I think you missed the specifier ‘commercial art’, i.e. art which no-one hugely cares for in the first place.

[Travis B.]

The thing, though, as we have repeated to you so many times, is that generative AIs simply can't replicate the abilities of human artists and writers. Consider the term "AI slop" ─ that is the crap that generative AI's churn out, and it is clear that it is obvious to the public that it simply does not match the output of human creators (or otherwise the term would not have been coined and would not have gained currency in the first place). People simply don't want what generative AI's produce, and while generative AI's may seem attractive to short-sighted executives who dream of replacing human workers, it is clear that generative AI's have a reverse Midas touch, and said executives will soon learn that when the content they sell suffers with the public.

Sure but what happens when AI improves to the point where it can compete with humans in terms of quality? You must understand that the computer industry is spending hundreds of billions every year on improving AI and it seems incredible that such investments will not yield noticeable advances eventually.

You're assuming that somehow generative AI will come to not be limited by its training sets, when in fact generative AI by its very nature is essentially an autocomplete that can only recreate what it has been trained with. The idea that it can do more than that is simply marketing by the techbros, which you seem to be taking at face value as the truth.

[WeepingElf]

AMEN!!! AI will perhaps soon dominate commercial artwork, but most people will insist on human-made art!

That statement contradicts itself though. If everyone insists on human-made art, then it hardly makes sense that AI comes to dominate art anyway. Unless you are suggesting that publishers and studios and so forth will simply ignore popular demand and force everyone to accept something they don't really want.

I think you missed the specifier ‘commercial art’, i.e. art which no-one hugely cares for in the first place.

Precisely that.

[malloc]

You're assuming that somehow generative AI will come to not be limited by its training sets, when in fact generative AI by its very nature is essentially an autocomplete that can only recreate what it has been trained with. The idea that it can do more than that is simply marketing by the techbros, which you seem to be taking at face value as the truth.

Sure but everyone outside the tech industry considered generative AI utter fantasy up until the moment it exploded on the scene. Almost everyone in 2021 would have dismissed LLMs and image generators as marketing hype with no chance of coming to fruition just as they dismiss further advances in AI now. Why were skeptics wrong then but correct now?

I think you missed the specifier ‘commercial art’, i.e. art which no-one hugely cares for in the first place.

Most people consume almost nothing but commercial art. Very few people are listening to nothing but live music from indie bands or reading nothing but zines and self-published novels. The vast majority of people watch television, play video games, and listen to music produced by massive media conglomerates. If all those corporations switch to AI over human artists, then the vast majority of media we consume will come from AI unless everyone makes a deliberate effort to boycott mainstream media.

[Raphael]

I think you missed the specifier ‘commercial art’, i.e. art which no-one hugely cares for in the first place.

That description only makes sense if you use a very unusual understanding of the term "no-one". (Perhaps as in, "No-one goes there any more. It's too crowded"?)

Commercial art is only commercially viable if enough people do care about it. Even though they might not be the kind of people who register as "people" to the Knowers of Art.

[Travis B.]

You're assuming that somehow generative AI will come to not be limited by its training sets, when in fact generative AI by its very nature is essentially an autocomplete that can only recreate what it has been trained with. The idea that it can do more than that is simply marketing by the techbros, which you seem to be taking at face value as the truth.

Sure but everyone outside the tech industry considered generative AI utter fantasy up until the moment it exploded on the scene. Almost everyone in 2021 would have dismissed LLMs and image generators as marketing hype with no chance of coming to fruition just as they dismiss further advances in AI now. Why were skeptics wrong then but correct now?

Because we understand how generative AI works and its fundamental limitations, which you seem to be willfully ignorant of in how you just unquestioningly buy into the techbros' marketing hype.

[WeepingElf]

You're assuming that somehow generative AI will come to not be limited by its training sets, when in fact generative AI by its very nature is essentially an autocomplete that can only recreate what it has been trained with. The idea that it can do more than that is simply marketing by the techbros, which you seem to be taking at face value as the truth.

Sure but everyone outside the tech industry considered generative AI utter fantasy up until the moment it exploded on the scene. Almost everyone in 2021 would have dismissed LLMs and image generators as marketing hype with no chance of coming to fruition just as they dismiss further advances in AI now. Why were skeptics wrong then but correct now?

Because we understand how generative AI works and its fundamental limitations, which you seem to be willfully ignorant of in how you just unquestioningly buy into the techbros' marketing hype.

Yes. And AFAIK those beasts still don't make any profit but burn a lot of money, so quite a few analysts expect the current AI hype to go bust soon. (However, autocrats may like AIs instead of human artists because AIs have no opinions of their own, and are thus easier to censor. But that still doesn't make an AI conspiracy of the kind malloc is fantasizing about.)

[Travis B.]

Sure but everyone outside the tech industry considered generative AI utter fantasy up until the moment it exploded on the scene. Almost everyone in 2021 would have dismissed LLMs and image generators as marketing hype with no chance of coming to fruition just as they dismiss further advances in AI now. Why were skeptics wrong then but correct now?

Because we understand how generative AI works and its fundamental limitations, which you seem to be willfully ignorant of in how you just unquestioningly buy into the techbros' marketing hype.

Yes. And AFAIK those beasts still don't make any profit but burn a lot of money, so quite a few analysts expect the current AI hype to go bust soon. (However, autocrats may like AIs instead of human artists because AIs have no opinions of their own, and are thus easier to censor. But that still doesn't make an AI conspiracy of the kind malloc is fantasizing about.)

Yeah, a big concern right now is about how we are now in a massive AI bubble which may pop at any moment, taking everyone's investments with it.

[malloc]

Yes. And AFAIK those beasts still don't make any profit but burn a lot of money, so quite a few analysts expect the current AI hype to go bust soon.

Why don't they make profit though? It seems like something that does the work of thousands of humans while costing less should yield amazing profit.

[Travis B.]

Yes. And AFAIK those beasts still don't make any profit but burn a lot of money, so quite a few analysts expect the current AI hype to go bust soon.

Why don't they make profit though? It seems like something that does the work of thousands of humans while costing less should yield amazing profit.

Do you realize how much energy and water and hardware generative AI takes? Do you realize what is entailed in training a generative AI model to do anything? These things cost significant quantities of money.

The only reason the AI companies haven't gone bust yet is that money to burn keeps getting shoveled in by investors caught up in the AI hype.

[WeepingElf]

Yes. And AFAIK those beasts still don't make any profit but burn a lot of money, so quite a few analysts expect the current AI hype to go bust soon.

Why don't they make profit though? It seems like something that does the work of thousands of humans while costing less should yield amazing profit.

Do you realize how much energy and water and hardware generative AI takes? Do you realize what is entailed in training a generative AI model to do anything? These things cost significant quantities of money.

The only reason the AI companies haven't gone bust yet is that money to burn keeps getting shoveled in by investors caught up in the AI hype.

Just that. While I expect the AI bubble to burst soon, I don't think generative AI will go away like a fashion fad. Such technologies tend to be here to stay, but the high expectations set in it will be moderated. So, people will do many things with generative AI, but it won't take over the world.

[Raphael]

While I expect the AI bubble to burst soon, I don't think generative AI will go away like a fashion fad. Such technologies tend to be here to stay, but the high expectations set in it will be moderated. So, people will do many things with generative AI, but it won't take over the world.

There once was an Internet bubble. It burst. But there is an Internet now.

[malloc]

Do you realize how much energy and water and hardware generative AI takes? Do you realize what is entailed in training a generative AI model to do anything? These things cost significant quantities of money.

The only reason the AI companies haven't gone bust yet is that money to burn keeps getting shoveled in by investors caught up in the AI hype.

Sure but one AI data center can produce as many images or lines of text as thousands of humans working around the clock. Does each data center really cost more than raising thousands of humans to adulthood and paying them to write or draw full-time?