Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

ChatGPT turned generative AI into an “anything tool”

arstechnica.com

Post reply on HN
Print view View original

121–130 of 245 posts

Re: ChatGPT turned generative AI into an “anything tool”

#121

Post by sillysaurusx »

ToucanLoucan wrote: post #119
Gigachad wrote: post #106

Earlier quoted context omitted.

I'm almost hoping for some enshittification. It's so depressing seeing how many skills and jobs have been instantly devalued. Take a look at the artist communities and it's total depression that peoples life passion and source of income has just been deleted over the course of a few months. We aren't talking about burger flipping and checkout scanners. This is a creative pursuit which takes years of grinding to becom…

Not even deleted! Actively stolen. No one has yet grappled with the reality that especially in the realm of generative art, these models do not work without a basis of absolutely stressful amounts of stolen artwork to train them. They simply do. Not. Exist. Without that baseline unethical contribution that tons of artists made without their knowledge and certainly without their consent. None of the major figures behi…

Oh please. First of all, how do you steal an idea? We’re talking about pictures. Supposing that you buy into the theory that you can, copyright was created to further the arts and sciences; it’s in the US constitution. The point isn’t to control your work — it’s to live in a richer society. And it’s not even clear that training a model counts as infringement. Being able to recite a quote from a book is different than reproducing the entire book. Artists won’t acknowledge that the same applies to their art.

If you believe that training models on art is stealing, then I’m a master ninja, since I’m the creator of books3. And even Stephen King today came out and said that he’s fine with it:

> Would I forbid the teaching (if that is the word) of my stories to computers? Not even if I could. I might as well be King Canute, forbidding the tide to come in. Or a Luddite trying to stop industrial progress by hammering a steam loom to pieces.

https://www.theatlantic.com/books/archive/2023/08/stephen-ki...

If he’s not worried, why are you?

I take a dim view of people trying to frame researchers as criminals. We’re not. We want to further science. That’s all.

You call me a grifter, but I’ve made roughly a hundred bucks from books3, and that’s because someone found my patron buried under a pile of links and subscribed to it many months ago. Most of my researcher colleagues seem to have similar distaste for wanting to make money. The work is the goal.

Re: ChatGPT turned generative AI into an “anything tool”

#122

Post by hansvm »

MPSimmons wrote: post #96
version_five wrote: post #75

Earlier quoted context omitted.

This is a good point and I'm surprised it hasn't had more attention. We've seen history repeat itself enough times that it's inevitable. Probably the best you can do is download a few of the full precision 70B or bigger models (LLaMA etc) to at least have some state of the art weights around you could use yourself if you wanted if/when openAI goes to shit. The nice thing about the language models is as least they're…

>... compared to google search which you'd never be able to host locally That's something I haven't thought about for a long time. How big is the Google index, would you say? Petabytes or exabytes, or am I vastly overestimating?

Probably petabytes for the text index and exabytes for the whole index (including images and video). The entire textual internet is hundreds of terabytes uncompressed, and indexes tend to balloon storage size to make serving tailored queries faster and cheaper. That idea shifts a bit with images/video where vector/ML techniques dominate; you get a reduction in quality, but quality is good enough (mostly, probably, we think), and you have within a small constant multiple one way or the other of the underlying dataset size for the index size (as opposed to text, where 100x-1000x one way or another isn't uncommon).

If the index is less than petabytes then it's impossible to serve many long tail queries even just based on text content. IME those queries _have_ gotten progressively worse results the last few years, so maybe the techniques have changed, but when the money printer is "search" and a petabyte of data is a fraction of a full-time engineer in cost to store, I doubt they would have cut costs that aggressively.

Re: ChatGPT turned generative AI into an “anything tool”

#123

Post by puchatek »

ToucanLoucan wrote: post #119
Gigachad wrote: post #106

Earlier quoted context omitted.

I'm almost hoping for some enshittification. It's so depressing seeing how many skills and jobs have been instantly devalued. Take a look at the artist communities and it's total depression that peoples life passion and source of income has just been deleted over the course of a few months. We aren't talking about burger flipping and checkout scanners. This is a creative pursuit which takes years of grinding to becom…

Not even deleted! Actively stolen. No one has yet grappled with the reality that especially in the realm of generative art, these models do not work without a basis of absolutely stressful amounts of stolen artwork to train them. They simply do. Not. Exist. Without that baseline unethical contribution that tons of artists made without their knowledge and certainly without their consent. None of the major figures behi…

Reminds me a bit of how everybody was cool with pirating software and media back in the day. To help the bad conscience there was a handy narrative of how the record companies were making a killing off of everybody so it was cool not to pay. But in reality it was just technically feasible. That's what enabled the behavior, not a few million Robin hoods coming together to do the right thing.

Similarly I feel creators need to make it impossible to have their work stolen for AI if possible. It will be tough though. They have half the world against them it seems.

Re: ChatGPT turned generative AI into an “anything tool”

#124

Post by asynchronous »

scarface_74 wrote: post #86
Someone1234 wrote: post #2

Although let's see if it is sustainable. I pay for ChatGPT Plus but most people aren't, and from what I've read they're losing money like crazy. We may wind up seeing this as the technical high point before the various competitors cut back on expensive hardware/energy usage for the free tier, and the quality of the responses takes a dive. This might sound a bit conspiratorial so apologies for that but back when the c…

> This might sound a bit conspiratorial so apologies for that but back when the competition for Google Assistant/Siri/Cortana/Alexa was super hot, the responses to the various voice assistance was almost eerie in that it could infer what you wanted/needed. Siri at least was always a simple intent based system that was never very intelligent and very much a brute voice method where someone went in and manually configu…

Reminds me of the memes/era when AI truly was just a million nested if statements

Re: ChatGPT turned generative AI into an “anything tool”

#125

Post by shriek »

Alupis wrote: post #70
consumer451 wrote: post #55

Earlier quoted context omitted.

> I'd like to have LawGPT I tried ChatGPT 4 for a specific TX county's laws. This was after I had spent hours doing traditional search. ChatGPT got everything correct, and even gave me some more direction.

> ChatGPT got everything correct My anecdote with technical information is exactly the opposite, unfortunately. I routinely use it to help guide my research - generally about topics I know something about but want to know more. It often will provide some response that is entirely wrong but looks very good. When called out, it apologizes, then proceeds to produce more information that's also not entirely correct. Ther…

Likewise, same result for me too but I get it. It's a language model. It makes a best guess based on the information it was provided so I trust it on the things I'm familiar with but verify on those that I'm not.

Re: ChatGPT turned generative AI into an “anything tool”

#126

Post by earth-adventure »

shriek wrote: post #125
Alupis wrote: post #70

Earlier quoted context omitted.

> ChatGPT got everything correct My anecdote with technical information is exactly the opposite, unfortunately. I routinely use it to help guide my research - generally about topics I know something about but want to know more. It often will provide some response that is entirely wrong but looks very good. When called out, it apologizes, then proceeds to produce more information that's also not entirely correct. Ther…

Likewise, same result for me too but I get it. It's a language model. It makes a best guess based on the information it was provided so I trust it on the things I'm familiar with but verify on those that I'm not.

I'm really stunned by the fact that, it seems, so few people get this. It's not a robot in the dense of Commander Data from Star Trek. It's a recursive software method that just has unfathomable amounts of data to base it's guess for some correct sentences on.

There is no logic, no reasoning, no nothing. Yet people complain it's 'giving me wrong answers'. Well, it doesn't know what's it's giving you either way. It only knows the statistics or odds about the sentences it creates being similar to what others have said before.

Re: ChatGPT turned generative AI into an “anything tool”

#127

Post by Gigachad »

kaashif wrote: post #116
Gigachad wrote: post #106

Earlier quoted context omitted.

I'm almost hoping for some enshittification. It's so depressing seeing how many skills and jobs have been instantly devalued. Take a look at the artist communities and it's total depression that peoples life passion and source of income has just been deleted over the course of a few months. We aren't talking about burger flipping and checkout scanners. This is a creative pursuit which takes years of grinding to becom…

If things slow down a bit then speed up again in a few years once economics improve or whatever, does that really make any difference in the long run? What's the long term solution here? We're not even talking about decades here, probably.

Part of me hopes we have hit a kind of wall where we get what we have now, maybe with some more polished UI, but it doesn't continue the exponential improvement path it's on. If it all slowed down, people would have plenty of time to readjust. Rather than just instantly deleting peoples careers.

Human artists can quite easily beat the current tech, but there is no real indication that the current stuff is the best we can juice out of the current methods. Which is the depressing part. Why even bother to learn anything when AI will obsolete your effort before you finish.

Re: ChatGPT turned generative AI into an “anything tool”

#128

Post by jdshaffer »

MPSimmons wrote: post #96
version_five wrote: post #75

Earlier quoted context omitted.

This is a good point and I'm surprised it hasn't had more attention. We've seen history repeat itself enough times that it's inevitable. Probably the best you can do is download a few of the full precision 70B or bigger models (LLaMA etc) to at least have some state of the art weights around you could use yourself if you wanted if/when openAI goes to shit. The nice thing about the language models is as least they're…

>... compared to google search which you'd never be able to host locally That's something I haven't thought about for a long time. How big is the Google index, would you say? Petabytes or exabytes, or am I vastly overestimating?

That reminds me of the time my friend downloaded the ENTIRE Yahoo! search database to a 1.44MB floppy. I think that was back in 1993. laugh

Re: ChatGPT turned generative AI into an “anything tool”

#130

Post by xigoi »

sillysaurusx wrote: post #121
ToucanLoucan wrote: post #119

Earlier quoted context omitted.

Not even deleted! Actively stolen. No one has yet grappled with the reality that especially in the realm of generative art, these models do not work without a basis of absolutely stressful amounts of stolen artwork to train them. They simply do. Not. Exist. Without that baseline unethical contribution that tons of artists made without their knowledge and certainly without their consent. None of the major figures behi…

Oh please. First of all, how do you steal an idea? We’re talking about pictures. Supposing that you buy into the theory that you can, copyright was created to further the arts and sciences; it’s in the US constitution. The point isn’t to control your work — it’s to live in a richer society. And it’s not even clear that training a model counts as infringement. Being able to recite a quote from a book is different than…

If OpenAI believes that copyright is wrong, why are they keeping their software proprietary?
Post reply on HN