Nick Clegg Says Asking Artists For Use Permission Would 'Kill' the AI Industry 218
As policy makers in the UK weigh how to regulate the AI industry, Nick Clegg, former UK deputy prime minister and former Meta executive, claimed a push for artist consent would "basically kill" the AI industry. From a report: Speaking at an event promoting his new book, Clegg said the creative community should have the right to opt out of having their work used to train AI models. But he claimed it wasn't feasible to ask for consent before ingesting their work first.
"I think the creative community wants to go a step further," Clegg said according to The Times. "Quite a lot of voices say, 'You can only train on my content, [if you] first ask.' And I have to say that strikes me as somewhat implausible because these systems train on vast amounts of data."
"I just don't know how you go around, asking everyone first. I just don't see how that would work," Clegg said. "And by the way if you did it in Britain and no one else did it, you would basically kill the AI industry in this country overnight."
"I think the creative community wants to go a step further," Clegg said according to The Times. "Quite a lot of voices say, 'You can only train on my content, [if you] first ask.' And I have to say that strikes me as somewhat implausible because these systems train on vast amounts of data."
"I just don't know how you go around, asking everyone first. I just don't see how that would work," Clegg said. "And by the way if you did it in Britain and no one else did it, you would basically kill the AI industry in this country overnight."
Nutshell (Score:5, Informative)
Obviously AI in its current incarnation is incapable of existing with consideration for artists/authors rights.
Re:Nutshell (Score:5, Insightful)
Of course. AI is not creative. It has to learn off existing material, be that text, voice, music, people's faces, paintings, drawings, television shows, anime, etc.
Like here's the thing. I would be OK with the AI scraping existing publicly-reachable information, if it only scraped it once and retained the credit. The problem is that it does neither of those. It does not respect the bandwidth websites pay for ( a problem that web crawlers/spiders have a problem with in the first place ) and it does not respect the copyright and ownership of the materials on the websites. Like that "studio ghibli" art style one of the AI's came out with... that is absolutely wrong. You should not be able tell the AI "studio ghibli" or name any of the films they were responsible for and get an art style like it. This is literately telling the AI to not be creative, but to clone the style.
And that's what AI is good at. Transforming thing A into thing B. It's not whole-cloth making anything. Music is the worst though because at present AI does not sing, either it can "choir" or it requires someone else to sing (eg the original artist) and squeeze it through an AI-autotune into another signer's voice style, but it's still very clearly the original singer.
Re: (Score:2)
Music is the worst though because at present AI does not sing, either it can "choir" or it requires someone else to sing (eg the original artist) and squeeze it through an AI-autotune into another signer's voice style, but it's still very clearly the original singer.
Man, you're behind the times a bit here. I've been playing around with Suno quite a bit and their latest model is so good it's creepy.
Hear it for yourself [youtu.be], that's one of the songs I made with the free trial of their paid model.
Now here's the thing, do I fancy myself an actual artist because I'd collaborated with ChatGPT to turn my ideas into some lyrics, then had Suno make it into a song? Not really, because being a "real" artist is about having the correct industry connections. Then you can get away wit
Re: (Score:2)
Re:Nutshell (Score:4, Insightful)
...but you could slip that into the playlist at Texas Roadhouse and no one would bat an eye.
That's because most human music is a regurgitation of existing music. We've been complaining about cookie-cutter music for decades now, and AI just confirms that there is no creativity in modern music. There are only so many chords (four, if I remember correctly) and progressions that sound good, and they can be (and have been) mathematically determined. AI is great at analyzing existing music and extracting the most used chords. It is, after all, a statistical analysis engine.
Meta and OpenAI keep confessing to massive copyright infringement, so prosecution should be a no-brainer.
In all copyright cases, the reproduction and/or integration of someone else's copyrighted material for commercial purposes require permission. AI must be opt-in. If AI dies for lack of training material, nothing of value will be lost.
Re: (Score:3)
We've been complaining about cookie-cutter music for decades now, and AI just confirms that there is no creativity in modern music.
And that's the status quo the copyright cabal is fighting tooth-and-nail to preserve. Songs such as Stupid Hoe [wikipedia.org], and Lift Yourself [wikipedia.org]. Lest anyone think I'm just singling out rap here, I'd also like to mention that even classics like Simon and Garfunkel's The Boxer [wikipedia.org] ended up with literal placeholder lyrics in the chorus. Of course, over time people have decided to attribute their own interpretation to give it deeper meaning, but it really was just a case of writer's block becoming a hit.
Finally, I'd like to a
Re: (Score:2)
Re: (Score:3)
Re: (Score:3)
Re: (Score:3)
He didn't really say that. The summary started out misrepresenting him. Then you get down to the last paragraph and find out what he actually said:
"And by the way if you did it in Britain and no one else did it, you would basically kill the AI industry in this country overnight."
That's probably true. If only one country had the restriction and no other, all the AI work would move to other countries. That's why we need international agreement that ignoring the rights of authors is not ok.
Re:Nutshell (Score:4, Informative)
The rights to control commercial reproduction of their works, primarily. Continental Europe mostly recognizes "moral rights" of authors and other creators that go beyond what US copyright law recognizes; I'm not sure where the UK falls on that topic.
Re: (Score:2)
Re:Nutshell (Score:5, Insightful)
AI doesn't consume concepts, styles or ideas. It consumes the specific instances of a creative work in its entirety, transforms it via an algorithm, and redistributes it as a commercial service for profit.
What part of that was copyright never intended to prevent?
Re:Nutshell (Score:5, Insightful)
The only difference between the output of a machine and a human, is that we can more accurately quantify how the "creative" process works when it's being performed by a machine.
What's being performed by a machine (the "AI") is not creative process. It's rearrangement of a lot of Star-Wars related input to produce some vaguely coherent Star Wars-related output. It's only able to do that because it hoovered up a lot of material of both the original Star Wars and related content. AI could not ever come up with Star Wars if they did not exist at that point.
And yes, humans can produce something just as derivative, mindlessly copying the original, as what AI produces. In fact, that's true for the vast majority of what humans produce. Now, something derivative can still have value if it's created with some thought behind it and brings maybe a new spin on existing ideas to the table, but AI is not capable of that. It can only copy others' homework without any understanding what any of it means. And we obviously want more completely original ideas.
With humans, even if 90% of what they produce is AI-level derivative (percentage is probably much higher but w/e), there is still the 10% that produce new ideas. Even if those, in their turn, will later launch a sea of derivative work, it keeps the culture from becoming completely stale. Unfortunately, for AI the percentage of derivative works is 100%, and that is not what we want.
Re: Nutshell (Score:3)
The only difference between
There are a TON of differences. Probably the biggest is that the machine version can read the entirety of all known creations.
Humans can study some a book in a few days, watch a movie in an hour or two, a web page in a few minutes. Machine learning can pull in thousands in the time it has taken you to read this.
Similarly for output, writing a book takes months to years, staging photos takes time and tools, feature films are hundred million dollar multi-year endeavors.
The human cost is a huge part of the
Re:Shill (Score:4, Insightful)
Shill? Nah, I'll straight up admit it, I don't work in a field that's likely to be automated away so to me this AI stuff is just a neat new toy to play with. I can't remember when /. got super cynical about new tech, but maybe it began with the iPod?
"No WiFi, less space than a Nomad, possibly will murder your children. Lame."
Close enough.
Re: (Score:3)
That is in fact what mimicry AIs are doing, quite blatantly. You can generate images with signatures from the works that were copied, without attribution. You can generate paragraphs from novels that were copied, without attribution.
By your own argument, AI companies should be fined for large scale commercial copyright infringement today. In fact, they are being sued for just that in multiple jurisdictions.
Re: (Score:3)
That is in fact what mimicry AIs are doing, quite blatantly. You can generate images with signatures from the works that were copied, without attribution. You can generate paragraphs from novels that were copied, without attribution.
There's this rather hilarious scene in Malcolm in the Middle [reddit.com] where Malcolm accidentally re-creates the Meow Mix commercial song.
Granted, yes, that's an example from Hollywood fiction, but it does actually happen with humans where something gets subconsciously plagiarized by mistake. Sometimes it's even on purpose, as was the situation with the Ghostbusters theme song, where the producers actually straight up prompted the AI song generator to re-create a song in the style of Huey Lewis. Oh right, they aske
Re: Nutshell (Score:5, Insightful)
Thats nonsense. Copyright was conceived on a fundamental level to protect and advance creative enterprise. It's not an archaic notion that need updating.
The real question is "should the creators of an LLM be allowed to profit on human creativity to the detriment of the humans doing the creating." Because that's what's happening. No LLM exists without human works. And most of the value comes from recent work. These kinds of things were fraught even when humans read a work and changed a few words to "transform" it, copyright has always been hard to quantify.
But it seems to me the line here is clear: if you use non licensed data the LLM cannot be monitized, and products of the LLM cannot enjoy copyright or be commercialized. Why should someone else profit on the backs of others without permission or compensation? And why should the weeping that "if you dont let us our business wont work!" be any more compelling than the guy selling unlicensed T-shirts outside a sports game?
And to the morons that say "well thats how humans learn why shouldn't an LLM do it" I say great! The LLM now owns it's outputs, no humans can. When you can convince humanity at large that the LLM is sentient and capable of licensing its own creativity you can purchase its output. Until then its a tool. And tools can't incorporate copyrighted material without licenses.
Re: (Score:3)
Thats nonsense. Copyright was conceived on a fundamental level to protect and advance creative enterprise.
Copyright was never intended to prevent competitive works from being created, nor was it intended to ensure all creative output has a market. If someone wants to buy Diddly's Adventures in Interstellar Conflict instead of Star Wars, Disney can go pound sand - that's perfectly legal. It also means if you want to enter the extremely crowded fray of mobile match-3 video games, you're free to do so, just don't expect a windfall in the process.
You're attributing aspects to copyright that weren't part of its sc
Re:Nutshell (Score:5, Interesting)
If the AI industry cannot deal with this, then what reason does it have to exist at all? I mean, if something relatively simple cannot be done, it shouldn't exist.
And if the industry keeps robbing in the ways they do, what incentive there is to still make (original) content? And if that lasts 5 years or more, the industry won't have anything new to learn and at that moment it also has no reason to exist. By robbing content creators blind, it is only postponing the inevitable, AI does not bring anything new/original/good to the table.
AI/LLMs are great at helping someone out making new code. or make transcriptions of meetings and such. It is poor at making new stuff. It is hyped up too much, there is too much money invested and the techbro's keep whipping up the frenzy about "we're falling behind....We're Falling Behind...WE'RE FALLING BEHIND!!!"
AI is a good tool with lots of potential. But it is managed by idiotic psychopaths, who have no other motive than to make away with as much money as they can, no matter the cost to any other company, people or society as a whole. Why do you think that OpenAI is slushing away so much funds from the investment money to buy John Ive's company/product that doesn't even exists?
Ive's is promised to have a large sum from that money when that venture fails and Altman will have the brunt of that money squirreled away, where only he can reach it. The rift between OpenAI and Microsoft is widening too. Altman is most definitely not the good guy he presents to the world in (video)interviews.
Deepseek R1 showed that the way forward with AI is to work smarter, not harder. But with the exodus of the "brains" at OpenAI, smarter is not an option, only harder. and wasn't the CEO from NVidia also saying that he will spend less on advancing AI hardware and invest more in robotics? That is also a sign that things are not as rosy in AI-land than most people believe.
And then you have a politician like Clegg who moans about AI, while being even further out of the loop than most AI companies and their people are. But his comments will be beneficial to them as the proverbial plebs have barely a clue at all about AI/LLMs.
The tool AI/LLMs and us, we certainly deserve better people at the helm of OpenAI, Anthropic, Google and Azure. So F it, locally hosted AI for the win, the rest can sink in an icy lake for all I care.
Re:Nutshell (Score:5, Insightful)
What rights would those be? The right to disallow someone to read their work?
The same ones programmers have to protect their copy and morals right to their code. Everyone fucks over Artists first because that set's the precedent to fuck over everyone else.
Have they assigned their right. No.
That's overly reductive (Score:5, Insightful)
That's pretty reductive. If that's all that was happening, then this conversation wouldn't be happening.
What is happening is that the work is being read, dissected down to its component word associations, then lifted wholesale and stored in an LLM database. Then it's being using to reproduce not only the artist's words, but the association patterns of word usage in order to duplicate the creator's own mental associations.
While fair use can cover quite a bit more than content providers want to admit, using a creator's work to duplicate the creator's thought process is where the stretch is. And they whine about when content creators complain. "We need to rip off their work or we can't rip off their thought process to duplicate", they cry.
Here's what needs to happen. The Berne convention needs a new clause that explicitly states that regardless of copyright status, regardless of whether a work is in the public domain or not, regardless of any other fair use, using any work as a source for LLM training is in contravention of copyright unless the creator has specifically and explicitly licensed it for use as training data.
Re: (Score:3)
Here's a thought: why not create an AI whose job is to be really good at tracing the copyright tree backwards to find the owner of any work, and ask the correct person for permission. Su
Is this bad? (Score:5, Insightful)
He says that as if it were a bad thing.
Re: Is this bad? (Score:4, Insightful)
Right? I mean, no other industry gets a pass for this behavior
Re: (Score:3)
Killing the AI industry in its current form _worldwide_ would be a good thing. A big reset and rethink, kind of like reining in the nuclear arms race. Killing the AI industry in the UK only, while it remaining a free-for-all elsewhere is economic harakiri. If the UK bans it, but some other country doesn't, then companies will simply go set up shop in that other country, do all the creative output mining where it's legal, and then sell whatever they can to whomever they can. They make money, the UK doesn't.
Re:Is this bad? (Score:5, Insightful)
I'd be OK if every AI crawler that dropped by a site put a nickel, several pence, or even bit into that site's coffer. Here's your dosh, here's my scrape.
They don't do that. It's kleptocracy, purely and simply.
With millions of BS sites on the web, AI is made up of millions of garbage pages masquerading as "intelligence". The quality of the web is hideous, and people scratch their heads when AI hallucinates.
The real and human content, no matter its quality, is as unrewarded as the BS goo found at every third IP address.
I hereby invent, AutoDosh. It's the tip jar for a crawler to get into a site. It has a unique code. If a training model doesn't like the content or it's redundant/useless drivel, remember not to go by that site and drop the nickel. Otherwise, gimme my nickels.
Re: (Score:2)
I would like to see the Google anti-trust trial solve this. A good solution is to have a single crawler for the web (can be Google's via anti-trust settlement) and then everyone pays into a pool to get access to the feeds from that single crawler. Payment into that pool can then be used to make the equivalent of statutory royalty payments to the sites crawled. If you don't want to be crawled put your stuff behind a login. Of course you are going to be sorely disappointed in the amount you get from those
Re: (Score:2)
Your model is good, save for the pool idea, as everyone will try to drain it inequitably. Direct payment. Kafka could handle it, or a pub/sub model funds disbursement model.
For those that have scraped prior to this, may they rot in hell, broke, with GPU payments to make.
Re: (Score:2)
The non-profit controlling the pool can negotiate the user fees based on usage. As for people who bypass the pooled crawler -- every web site should let them make requests and then never respond to those request effectively keeping them in infinite timeouts. Public embarrassment of the bypassing entities will also help control this.
Re: (Score:2)
I don't know if you've looked recently, but there is no such thing as embarrassment in all vectors of AI. But I get your idea.
I suggest also adding tokens by PAYING IN ADVANCE in some negotiated HARD CURRENCY to fuel the pool. Admittance is then a matter of a valid public key generated against the pool and site.
Re: (Score:2)
Now get the judge in the Google anti-trust trial to order Google to create this are you're done. I don't think the proposal is unreasonable, someone just needs to get it in front of the right people. Google would certainly prefer doing this instead of selling off Chrome.
Re: (Score:2)
Another option would be to let each page set a micro payment amount in it's headers. Then the crawler could crawl until they run out of money. This works as a double-edged sword. If you set your micro payment amount too high you are not going to get crawled and then you'll drop out of every search index. So it's your choice. The single crawler would crawl free pages first and then crawl from cheapest to most expensive until it runs out of money. Obviously if you set your micro payment at $100 you're never
Re: (Score:2)
There is a good solution to prevent gaming this. The crawler can use AI to assess whether it wants to pay the price the page is asking or not. It can always decide the price is to high and not add the page to the index. In that case it doesn't pay. The payment is not for crawling, it is for giving permission to be added to the global index.
Re: (Score:2)
You're both setting up curation. This Darwinian approach might also eliminate AI sucker-bait. I like the idea.
Translation (Score:3)
"We're the new hotness that everyone wants to throw huge stacks of cash at and use as an excuse for / against anything at all, but we don't want to spend any of those billions of dollars fairly compensating the creator of the works we want to create derivative works of for their copyright, which is explicitly under the control of the copyright holder."
Or, the TL;DR version: "waaaahh we don't want to pay licensing for their work, but expect everyone to pay us to license the derived product of their work"
And
Re:Is this bad? (Score:5, Insightful)
Year by year "right to a profitable business model" (as long as you have a big enough pile of wealth to begin with) marches onward.
When your businesses' "one neat trick" is *ignore the law* (because money), your business should not be permitted to exist.
Re: Is this bad? (Score:2)
Re: (Score:3)
"you would basically kill the AI industry in this country overnight."
He says that as if it were a bad thing.
The genie ain't going back in the bottle.
Re: (Score:2)
farnsworth_good_news.gif
Admission of guilt. (Score:5, Insightful)
If your industry is unable to survive by following the law then isn't that the same as admitting the basis of your industry is violating the law?
Re: (Score:2)
Not once you understand the Golden Rule: he who has the gold, rules.
Re: Admission of guilt. (Score:2)
Yep. What he's saying is the very definition of corporate theft.
Re: (Score:2)
It's a terrible example because I hate those damn things, but the dockless scooter rental industry followed a similar playbook.
The only good to come of it was that, in most places, the revised laws that resulted made it legal for privately-owned scooters to be ridden as a means of transport. Whereas previously, they existed in sort of a legal gray area as something you were technically only supposed to ride on private property.
IMHO, copyright lasts too long and it is badly in need of reform anyway. If the
Not a new thought. (Score:2)
Everyone is subject to all laws of all countries at all times. It is only when a nation chooses to limit the bounds of it's power that you are not bound by their laws. Therefore, the only limitation of another nation to enforce it's laws upon you is your physical location. To this end, many countries have extradition treaties.
To answer your question, how and when the law is enforced (contradictory or not) is up to that nation. An example of this is China which regularly uses their legal authority (in China)
Familiar argument (Score:5, Insightful)
That's the same argument Napster used.
If we can't get all our inputs for free, it would kill our ability to charge for similar stuff based on those inputs.
If we can't take everything we want without permission whenever we want it, it would kill our plan to replace human creativity with cheap imitations.
Re: (Score:2)
Shutting down Napster didn't kill music streaming, either. The industry went back to the drawing board, came up with a better business model, and now rules the entertainment world with it.
Re: (Score:3)
And that's exactly what is expected of AI proponents too. Cut them off from the free lunch. The alternative is the scrapping of copyright.
Hehe, my Slashdot CAPTCHA word is "plunders". So apt.
Re: (Score:2)
Re: (Score:3)
That's the same argument Napster used.
If we can't get all our inputs for free, it would kill our ability to charge for similar stuff based on those inputs.
Napster A. didn't charge, and B. didn't provide "similar stuff"; it provided identical stuff (ignoring compression artifacts). So no, that's not the argument Napster used.
Re: (Score:2)
Napster's argument was that they weren't breaking the law because they didn't host any of the content that was being illegally distributed. Technically that was true, but then contributory infringement became a thing.
AI isn't helping you get a bootleg copy of Taylor Swift's latest album. Well, it might explain the process with the right sort of prompt, but you'll still have to do the actual legwork yourself.
Re: (Score:2)
I fail.... (Score:2)
Umm, the point? (Score:2)
Asking Artists For Use Permission Would 'Kill' the AI Industry
First, a lot of hyperbole there (like AI couldn't possibly do anything else?!?), but, when you think of it, what's the alternative? Not asking artists, and kill off that industry (which, btw, if didn't exist these AI companies wouldn't have anything for their models to train on)
OK, so that in case (Score:3)
Make it mandatory, but force the AI industry to pay for every work they use. Photos and pictures could cost a million quid per photo/picture. And written material a thousand pounds per word.
Oh, and add an extra 10% on top of that to be redistributed as a UBI to the entire country.
If you're going to replace good quality work with slop, you need to compensate the people who'd have otherwise been able to get quality content. And if you're going to automate people's jobs out of existence and demand their labor be used to automate it, you should set them up for life.
Oh, your paymasters at Facebook don't like that Clegg you big fucking corporate shit? Maybe you should rethink your world view then.
Imagine one day (Score:2)
Seeing a computer on a street corner with a sign that says "starving AI"
Business plan (Score:3)
The business plan therefore is to grab content for free, digest it, then provide the resulting service for a fee?
Or pay for the input content and charge more/a lot for the service?
In other words, the business plan doesn't hold water so the only option is to lobby government and public opinion.
Sad.
You say that as though it's a bad thing (Score:2)
That strikes me as a feature, not a bug.
Furthermore, who do you think you are, that you imagine the Silly Valley cunts you just finished hanging with have some kind of moral right to do whatever the hell they want with artists' works? Un-fucking-believable.
Can we break out the torches and pitchforks yet?
Re: (Score:2)
"Artists" I'd empathize with. But we're talking about media conglomerates who just hoard IP like the proverbial dragon sitting on its pile of golden treasure.
I'm totally fine with slaying the dragon.
Don't kill the dead baby industry (Score:2)
Obviously the dead baby industry is going to suffer if I am not allowed to go around murdering babies without permission. You'll kill the dead baby industry if you continue down this path of enforcing laws against murdering babies.
Fine... (Score:2)
But first, go ask disney to use all their images for free to train your AI.
Bring all your important personal as well, and go unarmed.
Re: (Score:2)
Thing is, some people are unironically suggesting exactly that. Let's let the companies who have locked up IP for generations continue to dictate how it can be used, what could possibly go wrong?
Perhaps, but ... (Score:2)
Stem cell research (Score:3)
Before biologists figured out how to "revive" pluripotency in adult stem cells - sometime in the 90's if memory serves - only embryonic stem cells could be use for research on stem cell therapy. And of course, if any therapy was to be discovered, only embryonic stem cells could be used for that also.
I remember back then, researchers were crying out that the potential of stem cell therapies was so great that it was worth harvesting embryos just to get the cell. That was obviously something of an ethical issue.
The pressure was huge to sidestep ethics and carry on with this wonderful new groundbreaking technology. Humanity couldn't pass up the opportunity to take advantage of this because of silly outdated ethical principles!
Well, the law stood firm, and it forced the researchers to fnd another way to get pluripotent stem cells from adult cells - something which is now commonplace: nobody harvests embryos anymore.
Same thing for AI: if it can't exist without stealing from everybody, then it shouldn't exist. Nobody should cave to the AI industry's demands, however much pressure they put on the legistlator. It will force AI professionals to resolve the theft problem and reinvent their industry to be compatible with common decency. Just like the stem cell industry did.
Caving to the AI industry's demands is the lazy way out of the problem, wouldn't foster innovation and would simply entrench theft.
China is stealing (Score:2)
So we must steal too?
Though if I recall correctly chatgpt was first and is US based.
Ok, fine. No problem. (Score:2)
The trained AI will be free, right?
Bummer (Score:2)
I guess AI isn't compatible with IP and capitalism. Time to shut it down.
Future class action lawsuit brewing (Score:2)
.
Maybe this is how copyright terms get rolled back (Score:5, Interesting)
Copyright Act [wikipedia.org] of 1790 – established U.S. copyright with term of 14 years with 14-year renewal
Copyright Act of 1831 – extended the term to 28 years with 14-year renewal
Copyright Act of 1909 – extended term to 28 years with 28-year renewal
Copyright Act of 1976 – extended term to either 75 years or the life of the author plus 50 years
Copyright Term Extension Act of 1998 – extended terms to 95/120 years or life plus 70 years
Now here we are and the business world is clamoring for AI. And a way to solve the bulk of the potential copyright claims is to reduce the term of copyright...
Cat Fight (Score:2)
The lifetime+75 years people are battling with the Transhumanist tech bros.
It's a shame the small current creators will be swept up in this but two groups who are parasitic on society are going to war, so all I need to do is find some low-carb popcorn.
and its impossible (Score:2)
cool, any bad sides? (Score:2)
That's a great idea (Score:2)
So that's the solution! (Score:2)
Great! Now we know how to kill the AI industry. So let's do it!
Crooks asking cops for permission (Score:2)
Will kill the Criminal Defense Lawyer industry....
Other industries... (Score:3, Insightful)
We see the same in other industries... making murder illegal has almost killed the asassination industry.
Re: (Score:2)
Impractical. There are too many bots stealing too much stuff.
I like the opt in idea. It would be hell to regulate given that no one believes in following the law until they get caught, but it is a start.
Re: (Score:2)
Shame they can't memorise textbooks in a millisecond or so and then access it through memory.
Or were you suggesting you pirate all your kids textbooks?
Re: (Score:2)
Shame they can't memorise textbooks in a millisecond or so and then access it through memory.
Machines do a lot of things faster than humans; that's kind of the point. Should we make cars illegal because no human can run that fast?
Re: (Score:2)
They didn't make cars illegal, but they've regulated them from her to hell and back, precisely because moving something that big and heavy that fast among the general public causes all kinds of problems if it were not.
From seatbelts, and airbags, to requiring operators to have training and permits, and limiting where they can be used, and what direction they can go, and where they can turn, and limiting how fast they can be used, etc, etc, etc.
Is your car example meant to imply that are in favor of regulati
Re: (Score:2)
Machines do a lot of things faster than humans; that's kind of the point. Should we make cars illegal because no human can run that fast?
No, but we have very strong restrictions on their use. Try driving your car at its top speed on a public road.
Or even if we ignore top speed, in England there are loads of urban areas with a 20mph speed limit - about as fast as a normal human can run.
Re: (Score:2)
They are asking for a law to make discovery easier, it's less a special protection and more a high level legal intervention in response to a high level disregard of the law.
Re: (Score:3)
In order to train an LLM, a copy is fed into the LLM (even if it isn't stored in the LLM, numerous copies are made for the purpose of trianing during training). The purpose of that copy is to enrich the owners of the LLM, which is a commercial purposel. Under what license was that copy made?
Those ephemeral copies are copies. There's lots of (stupid and awful) legal precedents around this stuff. Console hackers and software modders get slapped with this stuff all the time. e.g. lawsuits that succeed with arg
Re:meanwhile... (Score:5, Insightful)
And the exact same thing can be said of more than a few human beings too.
Humans are special.
But no one is suggesting that people be forbidden to read and learn the plots, locations, background details, and characters of those books.
Agreed. The contents of your mind are beyond the reach of copyright.
But the contents of your computer memory and hard drives are not.
Memorize it all you want, but copies on your computer are legally different.
Nor is is it being suggested to forbid people from showing off or passing on their knowledge.
Well, yes, people ARE forbidden from doing exactly that.
For example, if you want to recite lord of the rings to a paying audience? you need license for that. Translate it to Klingon and distribute copies?... you need a license for that too.
Sure you can show off to friends in a bar or around the kitchen table all you like privately. That's all fair use. But try to commercialize it and distribute it, broadcast it, or perform it? Yes, people also need a license for that.
AI companies by and large are for profit enterprises copying data that doesn't belong to them and exploiting it for commercial gain by training LLMs on it. LLMs aren't people. It takes input, it runs an algorithm, it generates output. For it to have a model of LotR in there, either the model is getting passed in the prompt the user gave it (it's not) or its getting in there via the training on the copy of LotR and everything else that was fed in (yep this one). So copies were made to train, and some sort of derivative copies are persisted inside the model data too... which then are commercially exploited.
And even if it did work exactly same way as a human brain worked it still wouldn't matter, because its still a computer, and memory and storage, and those things are all treated by copyright law differently from a human mind.
Re:meanwhile... (Score:4, Insightful)
The trouble is that AI plagiarising is kind of like money laundering. You can't take the output of an AI and work out what went in, in general, an more than you can take the output of a hash function and determine the input that produced it. Reaching the bar necessary to win a lawsuit is probably impractical, and the AI companies and their expensive teams of lawyers know this.
Re: (Score:2)
Depending on the legal system, you don't have to prove it, you just make it plausible enough for a judge to force the AI companies to provide you with the evidence.
NYT got OpenAI's training set. If they win at the Supreme Court, every other registered copyright owner will have an easy job doing likewise for every major AI company in the US. It might be harder in the UK, dunno.
Re: (Score:2)
Derivative works are also controlled through copyright. Are you arguing that an AI trained on a specific copyrighted material isn't a derivative work?
Re: (Score:2)
My kids are training on their content for free at school.
You have a very odd definition of "free" going on here. Because apparently you think that textbooks are delivered to the classroom by the magical textbook fairy.
Here's a quick hint:
1. citizens pay taxes.
2. some of those tax revenues go to public schools.
3. public schools BUY the textbooks with tax revenue, from vendors that specifically resell textbooks to school districts with the intent to teach students from them.
Feel free to swap out "textbooks" with "library books" if you like. They are still license
Re: meanwhile... (Score:2)
That's the usual definition, he means that his kids aren't paying tuition and they aren't buying the textbooks themselves.
The fact that libertarians don't like paying taxes doesn't change what people commonly mean when they use the phrase "for free".
Re: meanwhile... (Score:2)
Re: meanwhile... (Score:2)
It's "copyright" for fuck's sake. Also it's means "it is".
Re: (Score:3)
It's "copyright" for fuck's sake. Also it's means "it is".
I have this strange urge to write a post about page tables and spell it "copy on right" and watch the chaos. But yes, this annoys me, too.
Re: (Score:2)
That's where you're wrong dude.
The AI is ingesting whole works, and in many cases, can spit out entire pieces of those works without citing anything.
Like the best thing AI can do going forward is have a special tag for the Author Copyright, and a special tag for the source, and then always cite the author copyright and and sources that generate every response to a prompt. For artwork this has to go one further and actually create it in the metadata.
Re: (Score:2)
"Quite a lot of voices say, 'You can only train on my content, [if you] first ask.' And I have to say that strikes me as somewhat implausible because these systems train on vast amounts of data."
Let's rephrase this to see if it makes sense.
"Quite a lot of voices say, 'You can only deport people if you first give them due process.' And I have to say that strikes me as somewhat implausible because there are a LOT of people we want to deport."
Noting that Trump said just that. From Trump wants to bypass immigration courts. Experts warn it's a 'slippery slope.' [npr.org]:
"I hope we get cooperation from the courts, because we have thousands of people that are ready to go out and you can't have a trial for all of these people," Trump told reporters in the Oval Office [youtube.com] last week [mid April].
Bottom line, there are things you're suppose to do even if they're inconvenient.
It's also a crock of s**t. When the web came into existence, folks came up with a robots.txt file that gave permission to spider the site and, to a limited extent, serve snippets of the content. There's nothing preventing an ai_robots.txt file with a standard format that dictates whether the content can be used for AI training, whether it can be used for training of commercially available AIs or just fully open source AIs, etc. It won't kill the industry, unless the assumption is that nearly everyone wil
Re: (Score:3)
When the web came into existence, folks came up with a robots.txt file that gave permission to spider the site and, to a limited extent, serve snippets of the content. There's nothing preventing an ai_robots.txt file ... Opt-out is reasonable, ...
Sure, but compliance is voluntary -- according to my understanding, and Wikipedia robots.txt [wikipedia.org]:
The standard, developed in 1994, relies on voluntary compliance. Malicious bots can use the file as a directory of which pages to visit, ...
Security:
Despite the use of the terms allow and disallow, the protocol is purely advisory and relies on the compliance of the web robot; it cannot enforce any of what is stated in the file.
So "opt-out" works, if searchers honor it.
Re: (Score:2)
Well, voluntary enforcement makes sense, because if you really don't want people reading your content, you can always put it behind a login or paywall.
Re: (Score:2)
Re: (Score:2)
When the web came into existence, folks came up with a robots.txt file that gave permission to spider the site and, to a limited extent, serve snippets of the content. There's nothing preventing an ai_robots.txt file ... Opt-out is reasonable, ...
Sure, but compliance is voluntary -- according to my understanding
Compliance being voluntary doesn't change anything. The web spider companies comply because it gives them a reasonable belief that their usage is authorized.
That said, if such a scheme existed for AI, you would have reasonable cause to sue companies that don't comply, as by making it crystal clear that their usage is NOT authorized, such crawling becomes prima facie illegal under the CFAA.
Re: What the Eyes see and the Ears hear... (Score:2)