Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
Tei
Feb 19, 2011

SaTaMaS posted:

Just the opposite, at the moment the AI-Math job market is saturated, while the app developers creating the apps which use AI APIs are harder to find.

cat botherer posted:

Math people are a dime a dozen, unfortunately. Most applied ML doesn't really touch that advanced statistics. Businesses generally don't care about statistical soundness in my experience (dumb, but statistics just aren't buzzy enough for managers). CI/CD people are doing relatively well, but so are programmers (in comparison to data scientists). Software engineering is actually harder that classic script-monkey data science, which is why a lot of the demand is moving to data/ML engineering type positions. Most businesses want somebody who understands the ML stuff enough but can actually integrate it into production.

Ha... I guess it makes sense. I guess my vision of the market is somewhat distorted.

Adbot
ADBOT LOVES YOU

SCheeseman
Apr 23, 2003

This seems significant
https://www.reuters.com/legal/litigation/judge-pares-down-artists-ai-copyright-lawsuit-against-midjourney-stability-ai-2023-10-30/

quote:

Oct 30 (Reuters) - A judge in California federal court on Monday trimmed a lawsuit by visual artists who accuse Stability AI, Midjourney and DeviantArt of misusing their copyrighted work in connection with the companies' generative artificial intelligence systems.

U.S. District Judge William Orrick dismissed some claims from the proposed class action brought by Sarah Andersen, Kelly McKernan and Karla Ortiz, including all of the allegations against Midjourney and DeviantArt. The judge said the artists could file an amended complaint against the two companies, whose systems utilize Stability's Stable Diffusion text-to-image technology.

Orrick also dismissed McKernan and Ortiz's copyright infringement claims entirely. The judge allowed Andersen to continue pursuing her key claim that Stability's alleged use of her work to train Stable Diffusion infringed her copyrights.

The same allegation is at the heart of other lawsuits brought by artists, authors and other copyright owners against generative AI companies.

"Even Stability recognizes that determination of the truth of these allegations – whether copying in violation of the Copyright Act occurred in the context of training Stable Diffusion or occurs when Stable Diffusion is run – cannot be resolved at this juncture," Orrick said.

The artists' attorneys Joseph Saveri and Matthew Butterick said in a statement that their "core claim" survived, and that they were confident that they could address the court's concerns about their other claims in an amended complaint to be filed next month.

A spokesperson for Stability declined to comment on the decision. Representatives for Midjourney and DeviantArt did not immediately respond to requests for comment.

The artists said in their January complaint that Stability used billions of images "scraped" from the internet, including theirs, without permission to teach Stable Diffusion to create its own images.

Orrick agreed with all three companies that the images the systems actually created likely did not infringe the artists' copyrights. He allowed the claims to be amended but said he was "not convinced" that allegations based on the systems' output could survive without showing that the images were substantially similar to the artists' work.

The judge also dismissed other claims from the artists, including that the companies violated their publicity rights and competed with them unfairly, with permission to refile.

Orrick dismissed McKernan and Ortiz's copyright claims because they had not registered their images with the U.S. Copyright Office, a requirement for bringing a copyright lawsuit.

The plaintiffs can amend and re-file but given that those claims don't hold up and they have to lean on their core claim, that means the defendants would probably go for the Google Books fair use defense, giving them a fairly good chance of winning.

Tree Reformat
Apr 2, 2022

by Fluffdaddy
That lawsuit was always the weakest of the bunch, tbf.

The fact they specifically included DeviantArt rather than OpenAI always felt like a rather petty and performative personal shot from (and for an audience of) explicitly pissed off digital artists to me.

But they can't really rally around Getty's lawsuit as much anymore since oops they're running their own GenAI now.

biznatchio
Mar 31, 2001


Buglord

SCheeseman posted:

This seems significant
The plaintiffs can amend and re-file but given that those claims don't hold up and they have to lean on their core claim, that means the defendants would probably go for the Google Books fair use defense, giving them a fairly good chance of winning.

A very important aspect of the ruling is that the judge dismissed several of the claims with prejudice (meaning they can't be amended and re-filed) because the artists did not have registered copyrights on works related to those claims at the time of the alleged infringement (when the models were trained).

This is important because under US Copyright law, even though you have inherent copyright over a work the instant it is created, that implicit inherent copyright only covers actual damages; you are only eligible for statutory damages for infringements that occur after a work is registered with the Copyright Office. The artists did not allege any actual harm, and so since those works were not registered at the time, there is no remedy the court can provide even if copyright infringement occurred during the model training. Even if they run out and register copyrights on the works right now, it's too late, the statutory protection is not retroactive; thus the dismissal with prejudice -- there is literally nothing the artists can do to salvage those particular claims.

They would need to file a substantially different claim, one where they allege and can prove actual damages (and that's an extremely difficult task given the nature of what AI does and its highly abstracted relationship to any specific training material); otherwise there is no reckoning for the models trained on their artwork -- nor for the mountains of other artwork with unregistered copyrights at the time the existing models were trained. It's as if a million DeviantArt members cried out and were suddenly silenced.

Monglo
Mar 19, 2015
Well, this is good news!

Tei
Feb 19, 2011

The hollywood writers strike ended. Among many things, they asked to never be asked to correct scrips made with AI, that they won, but failed to get the one where they asked their script except from being feed to AI's.

Imaginary Friend
Jan 27, 2010

Your Best Friend
Are there any laws for the individual generating made with potential infringement of copyrighted material? I mean, artists constantly sue each other for stealing. Could this artist can sue an individual generating things that the AI obviously took inspiration from? Or does this turn into another weird grey zone where "nobody" actually created the art?

Tree Reformat
Apr 2, 2022

by Fluffdaddy

Tei posted:

The hollywood writers strike ended. Among many things, they asked to never be asked to correct scrips made with AI, that they won, but failed to get the one where they asked their script except from being feed to AI's.

The AI stuff for the writers' specifically always felt more like an ancillary bargaining chip to me compared to the residuals and writers' rooms stuff. I think the writers got exactly the compromise they wanted: they can't be forced to use or edit it, but they are free to use it if they want to. I think that's become a bone of contention now (like with novel writers using generated cover art) with other creatives. The actors strike, for example, is continuing partially because GenAI is more immediately an existential issue for them than it was perceived as for the writers (and apparently the directors union leadership).

Imaginary Friend posted:

Are there any laws for the individual generating made with potential infringement of copyrighted material? I mean, artists constantly sue each other for stealing. Could this artist can sue an individual generating things that the AI obviously took inspiration from? Or does this turn into another weird grey zone where "nobody" actually created the art?

The most US law currently states is a ruling from the US Copyright Office which uses the precedent from the Naruto case to declare generated material ineligible for copyright, since a human didn't produce it, although compositions that incorporate generated material can still be copyrighted for the parts the human did make. Edits of generated material have not yet been legally tested.

Most lawsuits are currently focused on the companies providing the GenAI tech rather than the users of it.

Imaginary Friend
Jan 27, 2010

Your Best Friend

Tree Reformat posted:

The most US law currently states is a ruling from the US Copyright Office which uses the precedent from the Naruto case to declare generated material ineligible for copyright, since a human didn't produce it, although compositions that incorporate generated material can still be copyrighted for the parts the human did make. Edits of generated material have not yet been legally tested.

Most lawsuits are currently focused on the companies providing the GenAI tech rather than the users of it.
Ah I meant the other way around. If a company or someone creates art using AI, and the generation they use has done a freaky generation where some part of the art piece is clearly grabbed from some real artists, does that artist have a case? Or if a company explicitly uses the style of an artist they maybe fired or whatever to create their future large franchise? I guess they might be owning the style as well when they make artists sign up for working with them.

Abhorrence
Feb 5, 2010

A love that crushes like a mace.

Imaginary Friend posted:

Ah I meant the other way around. If a company or someone creates art using AI, and the generation they use has done a freaky generation where some part of the art piece is clearly grabbed from some real artists, does that artist have a case? Or if a company explicitly uses the style of an artist they maybe fired or whatever to create their future large franchise? I guess they might be owning the style as well when they make artists sign up for working with them.

Styles aren't copywriteable.



Tree Reformat posted:


The most US law currently states is a ruling from the US Copyright Office which uses the precedent from the Naruto case to declare generated material ineligible for copyright, since a human didn't produce it, although compositions that incorporate generated material can still be copyrighted for the parts the human did make. Edits of generated material have not yet been legally tested.

Most lawsuits are currently focused on the companies providing the GenAI tech rather than the users of it.

Worth noting that this case should probably not be treated as some strong precedent, because the person who tried to file the copyright did everything in their power to minimize and downplay human involvement.

BrainDance
May 8, 2007

Disco all night long!

Abhorrence posted:

Worth noting that this case should probably not be treated as some strong precedent, because the person who tried to file the copyright did everything in their power to minimize and downplay human involvement.

Not just downplay, refuse it. But it wasn't a real attempt at copyright but one of those "force them to decide on things they otherwise wouldn't decide on" things


It was widely misreported in a really clickbait kinda way too, lots of "judge rules AI art can't be copyrighted" when, leading up to it (the case went through a lot of different stages) they were pretty explicit saying "we would look into whether or not this has enough human involvement to be copyrightable, but Thaler specifically states he is not the author and that he wants the AI to have the copyright, and non-human things can't hold copyright on anything"

It ruled all AI art can't be copyrighted in the same way the monkey selfie case ruled all photography can't be copyrighted. It affected humans trying to copyright AI generated stuff the same way the monkey selfie case affected human photographers too. People aren't trying to get a copyright under the name "Midjourney"

There was the Zarya of the Dawn thing that did look into how much human involvement is needed, and they weren't going to give a copyright to at least straight from midjourney stuff though. That's much, much more relevant than the Thaler case. That decision wasn't based on the monkey selfie stuff, though.

BrainDance fucked around with this message at 05:24 on Nov 3, 2023

biznatchio
Mar 31, 2001


Buglord
Basically in order for something to even be copyrightable in the first place there needs to be some level of human creativity involved in the creation of the work. It doesn't have to be a lot. Books of sports statistics can be copyrighted even though they're little more than tables of facts, the facts are copyrightable but the arrangement of them into a table is. And that's a lot less creative effort than some of the advanced work some people are doing to guide an AI into generating what they want.

The debate gets obscured because most people's exposure to AI is just a simple textbox you type some descriptive words into and you get something out the other end; but the real wizards are building processing flowcharts and masking images and drawing pose guides; and there's no way to say there's not human creativity in that.

The debate also gets obscured by artists who are very loud about not liking AI and reflexively always staking out the most anti-AI position they can on every issue possible out of spite and end up cornered into the argument that what they do is somehow uniquely special and wrong to automate through technology as compared to everyone else in the long line of professions who've already had computers come in and disrupt what they do.

Level heads will eventually come up with some form of copyright protection over AI-generated works. The current grey state with a needle pointed weakly to "not copyrightable" won't be the case for long.

Tei
Feb 19, 2011

Abhorrence posted:

Styles aren't copywriteable.

Thieves hidding in the technicalities of the law. Taking people work, then reselling that work in a way that with the current law is not illegal*.


* so far, AFAIK, pending judgements.

Monglo
Mar 19, 2015
Apparently The Beatles have one of their songs released, cleaned up by AI. Im seeing people burst their vessels calling this cultural necrophilia. disgusting, etc. how they hate it and doesnt sound anything like the Beatles. They think the song is fully written by AI

lol

Owling Howl
Jul 17, 2019

Tei posted:

Thieves hidding in the technicalities of the law. Taking people work, then reselling that work in a way that with the current law is not illegal*.


* so far, AFAIK, pending judgements.

Well it would generally speaking make art a complicated affair if you could copyright styles. Pretty much no matter what kind of painting, movie or music you could conceive of someone could claim it is in the style of something that came before it.

Tei
Feb 19, 2011

Owling Howl posted:

Well it would generally speaking make art a complicated affair if you could copyright styles. Pretty much no matter what kind of painting, movie or music you could conceive of someone could claim it is in the style of something that came before it.

you are correct. imo.

110723_8
Nov 8, 2023
and this is why siri doesn't work

Roadie
Jun 30, 2013

Monglo posted:

Apparently The Beatles have one of their songs released, cleaned up by AI. Im seeing people burst their vessels calling this cultural necrophilia. disgusting, etc. how they hate it and doesnt sound anything like the Beatles. They think the song is fully written by AI

lol

It sounds like the most boring parts of the Beatles' work, which is why it was never used or rerecorded before now. They just ran out of any other old material.

fart blood
Sep 13, 2008

by VideoGames
Sam Altman got fired

City Slicker
May 28, 2020

fart blood posted:

Sam Altman got fired

Maybe something to do with the allegations his sister was making?

They were brushed under the carpet a while ago, but seemed to have been getting traction recently.

fart blood
Sep 13, 2008

by VideoGames

City Slicker posted:

Maybe something to do with the allegations his sister was making?

They were brushed under the carpet a while ago, but seemed to have been getting traction recently.



Uh oh. Ouch. gently caress him if that’s true.

BrainDance
May 8, 2007

Disco all night long!

OpenAI just got hit with a bunch of downtime/instability in the last week or so, plus them having to stop new ChatGPT Plus subscriptions, it seems like things have been kind of a clusterfuck over there recently.

And with this
“Mr. Altman’s departure follows a deliberative review process by the board, which concluded that he was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities,”

I dunno, I don't think it was the sister thing, even though who knows what exactly it was. It could have been something like him lying to the board about financial stuff, or something that pushed them into having to make bad decisions like the ChatGPT Plus thing (which company wants to have to say "Sorry, for this part of our business, no one can give us more money"? Though I do wonder what their income from ChatGPT Plus looks like compared to the API.

Tei
Feb 19, 2011

USA uses his powerful financial muscle to run tech companies at a loss for a very long time. Until they dominate the world market. Then they flip the switch and raise prices/limit the service friendlines.

But I wonder if ChatGPT can ever turn a profit, if ChatGPT Plus does not and is already expensive. How much it need to ask users to pay for the servers and make a profit? 50$/month?

Owling Howl
Jul 17, 2019

Tei posted:

USA uses his powerful financial muscle to run tech companies at a loss for a very long time. Until they dominate the world market. Then they flip the switch and raise prices/limit the service friendlines.

But I wonder if ChatGPT can ever turn a profit, if ChatGPT Plus does not and is already expensive. How much it need to ask users to pay for the servers and make a profit? 50$/month?

Right now there's millions of people using ChatGPT as a toy to play with and it's probably too expensive for that to continue. It's also not really something the average person needs. It's like photoshop et al in that it will be indispensable and integral to the work and hobbies for some people while most people wouldn't immediately notice if it suddenly ceased to exist.

I'm guessing we'll see some big shiny models that will cater to the corporate market and a lot of smaller projects of varying capabilites from hobbyists and smaller companies. I suppose there could be one model that corners the corporate market but I suspect it doesn't benefit from the size of the user base in the same ways search, operating systems or social media does for differing reasons.

cat botherer
Jan 6, 2022

I am interested in most phases of data processing.
Everyone is rushing to get their customer-service chatbots out. In a vacuum, that could be worthwhile compared to paying call center employees, but I think the shine will wear off quick. They don't work particularly better than previous generations of chatbots for that purpose, and customers are extremely hostile to it.

SaTaMaS
Apr 18, 2003
https://arstechnica.com/information-technology/2023/11/report-sutskever-led-board-coup-at-openai-that-ousted-altman-over-ai-safety-concerns/

Wouldn't have thought of Ilya Sutskever as the John Sculley type but it sounds like Sutskever thought Altman was moving too fast and letting his ego get in the way of "AI safety". The astonishing thing is that Microsoft had no idea this was coming, and given that a half dozen top researchers are leaving with Altman, Sutskever had better have really thought this through.

Staluigi
Jun 22, 2021

cat botherer posted:

Everyone is rushing to get their customer-service chatbots out. In a vacuum, that could be worthwhile compared to paying call center employees, but I think the shine will wear off quick. They don't work particularly better than previous generations of chatbots for that purpose, and customers are extremely hostile to it.

Sorry, i did not find any good matches for "cancel my subscription"

Can we recommend upgrading to platinum service tier?

Nervous
Jan 25, 2005

Why, hello, my little slice of pecan pie.

Staluigi posted:

Sorry, i did not find any good matches for "cancel my subscription"

Can we recommend upgrading to platinum service tier?

Platinum service members are eligible to use our premium customer support service, connecting you with a live* rep!

*rep may still be a GPT rep depending on support workload**

**rep will still always be a GPT rep***

***gently caress you you asked for this

Nervous
Jan 25, 2005

Why, hello, my little slice of pecan pie.
Looks like Sam might be coming back and the board might have spooked after all the resignations.

https://www.theverge.com/2023/11/18/23967199/breaking-openai-board-in-discussions-with-sam-altman-to-return-as-ceo

BoldFace
Feb 28, 2011
I want to see all OpenAI employees quit and join Altman's new company just because how hilariously catastrophic it would be to Microsoft.

LionArcher
Mar 29, 2010


biznatchio posted:

Basically in order for something to even be copyrightable in the first place there needs to be some level of human creativity involved in the creation of the work. It doesn't have to be a lot. Books of sports statistics can be copyrighted even though they're little more than tables of facts, the facts are copyrightable but the arrangement of them into a table is. And that's a lot less creative effort than some of the advanced work some people are doing to guide an AI into generating what they want.

The debate gets obscured because most people's exposure to AI is just a simple textbox you type some descriptive words into and you get something out the other end; but the real wizards are building processing flowcharts and masking images and drawing pose guides; and there's no way to say there's not human creativity in that.

The debate also gets obscured by artists who are very loud about not liking AI and reflexively always staking out the most anti-AI position they can on every issue possible out of spite and end up cornered into the argument that what they do is somehow uniquely special and wrong to automate through technology as compared to everyone else in the long line of professions who've already had computers come in and disrupt what they do.

Level heads will eventually come up with some form of copyright protection over AI-generated works. The current grey state with a needle pointed weakly to "not copyrightable" won't be the case for long.

Exactly. I am an indie writer. I want my cover art to have AI elements. I want to prompt write a drawn body, tweak it, and then put a models face from deposit photo that I bought the rights too, then blend it/paint it using photoshop/AI to make a sick cover. That is me being very hands on and creative in the process. Saying I’m “stealing” from an artist because I do all that is… silly.

111923
Nov 19, 2023
structurally unable to exist renew my subscriptions due to Criminals and a beyond recognition weak Community

Serotoning
Sep 14, 2010

D&D: HASBARA SQUAD
HANG 'EM HIGH


We're fighting human animals and we act accordingly

Owling Howl posted:

Right now there's millions of people using ChatGPT as a toy to play with and it's probably too expensive for that to continue. It's also not really something the average person needs. It's like photoshop et al in that it will be indispensable and integral to the work and hobbies for some people while most people wouldn't immediately notice if it suddenly ceased to exist.

I'm guessing we'll see some big shiny models that will cater to the corporate market and a lot of smaller projects of varying capabilites from hobbyists and smaller companies. I suppose there could be one model that corners the corporate market but I suspect it doesn't benefit from the size of the user base in the same ways search, operating systems or social media does for differing reasons.

Ehhh I dunno, part of the inherent value/appeal of ChatGPT and similar models is that they are very accessible and require little to no expertise to get quite a bit of value out of. You just describe in plain language what you want it to do or fetch and it does it for you. With some tweaking they can easily become the "search engines" of the future. Microsoft is certainly betting on that angle early with Copilot.

BoldFace
Feb 28, 2011
Looks like Sam is not coming back after all.

https://twitter.com/_akhaliq/status/1726467905527611441

Original_Z
Jun 14, 2005
Z so good
https://twitter.com/satyanadella/status/1726509045803336122

Somehow I don’t think he’ll have to worry about putting profits over all anymore!

BoldFace
Feb 28, 2011
There's a huge flurry of OpenAI employees posting "OpenAI is nothing without its people" on Twitter. I don't know if it's just a call for the OAI board to resign or a sign that they're willing to follow Sam over to Microsoft.

cat botherer
Jan 6, 2022

I am interested in most phases of data processing.

Pvt. Parts posted:

Ehhh I dunno, part of the inherent value/appeal of ChatGPT and similar models is that they are very accessible and require little to no expertise to get quite a bit of value out of. You just describe in plain language what you want it to do or fetch and it does it for you. With some tweaking they can easily become the "search engines" of the future. Microsoft is certainly betting on that angle early with Copilot.
It's too expensive to run to ever be profitable as a search engine.

Mega Comrade
Apr 22, 2004

Listen buddy, we all got problems!
Has bing ever been profitable? A bing AI search is as much as 10x as expensive to do.

How does that model get to profitability?

Xand_Man
Mar 2, 2004

If what you say is true
Wutang might be dangerous


If you're trying for the enshittification pipeline it's a VC's wetdream. Forget Google, people will go to ChatGPT for everything

Then you jack up the prices or insert a paragraph about the great taste of Pepsi into their book report about Jonny Tremaine

Adbot
ADBOT LOVES YOU

Tei
Feb 19, 2011

Hope corporations lose control of the AI.

One way I could see this fit into corporations strategy is somebody rent you the server to run ChatGPT, and somebody else sell your the trained dataset. And everything else is open source. So corpos make money from the dataset and renting the server, and everything else is just a commodity.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply