We present Audio Flamingo 3 (AF3), a fully open state-of-the-art (SOTA) large audio-language model that advances reasoning and understanding across speech, sound, and music. AF3 introduces: (i) AF-Whisper, a unified audio encoder trained using a novel strategy for joint representation learning across all 3 modalities of speech, sound, and music; (ii) flexible, on-demand thinking, allowing the model to do chain-of-thought-type reasoning before answering; (iii) multi-turn, multi-audio chat; (iv) long audio understanding and reasoning (including speech) up to 10 minutes; and (v) voice-to-voice interaction. To enable these capabilities, we propose several large-scale training datasets curated using novel strategies, including AudioSkills-XL, LongAudio-XL, AF-Think, and AF-Chat, and train AF3 with a novel five-stage curriculum-based training strategy. Trained on only open-source audio data, AF3 achieves new SOTA results on over 20+ (long) audio understanding and reasoning benchmarks, surpassing both open-weight and closed-source models trained on much larger datasets.
I don’t think my opinion as some random dude matters here. I could uphold arbitrary stupid believes. But this is kind of a factual question. So whether I personally, as one person, am fine with something is of no concern here. The question is, how do we arrive at a consistent economy model for immaterial goods…
And I think I wrote like 5 times now that I’m NOT fine with that. I said I view it as a (necessary) evil. It is evil in the sense of bad, I’m not fine with it, it comes with severe issues, we should do better than that. However “is” and “should” are two seperate things. We happen to live on a world that came up with copyright. It exists. We made a pact with the devil to address one thing. And I’m merely acknowledging that. Since it does exist, I need to deal with it. That’s not agreement from my side. Copyright serves one legitimate purpose. It applies our capitalist economy to immaterial goods. It’s supposed to allow individuals and companies to create, and trade with more than just cocoa beans. But it’s complicated and we might have come up with a stupid way to do it. And a way that simultaneously has lots of negative side-effects.
And now what? That is the question. Do we abolish it? Do we replace it with something else that handles the one legitimate purpose a better way? Do we retrofit it and try to “patch” it? Do we do that just for AI? Or for more than just one use-case?
And I think I make a point about how return on investment and an economic rent are two distinct things. Yet they’re in practice falsely(!) mushed together, which again is bad… Or am I mistaken and I can pay an artist for their investment but not pay a rent? I don’t think there is a good way to do it with the current model. That means I get to treat both as the same. You seem to be under the impression I like it. But I don’t. It’s just that I have to abide by law and that currently mandates me to do it.
Or am I mistaken and I can pay an artist for their investment but not pay a rent?
A return on investment is not the same as an economic rent.
Let’s go back to the farmer example. You agree that a monopoly on the food supply is a bad thing. It can and will be abused.
Sidenote: You suggested that the government should produce textbooks to prevent abuse. Would that also be your solution here? Would that be preferable to the current arrangement?
Now, let’s look at the situation of a farmer more closely. A farmer has to do a lot of work before they can harvest. They also need stuff like seeds, fertilizer, pesticides, fuel, machinery, spare parts and maintenance, and so on.
In the old times, one held back part of a grain harvest as seed grain for next year. That is an investment in the economics sense. You don’t consume everything, but keep it so that you have more in the future. The finance meaning is subtly different but never mind.
Farmers gets a return on investment. They invest money and labor so that there is a harvest in the future. They could sell the equipment they already own to have more spending money now.
A ROI is part of a farmers’ income but is not economic rent.
Back to authors. An established author will get an advance before they write the next book. That’s investment by the publisher. If they don’t get an advance, then the author is making the investment, but let’s ignore that for simplicity. Investments are always risky. In this case, some books don’t sell well and don’t make back the money.
As a publisher, how much money would you invest in future books to maximize your profit? It depends on the expected payout and the cost of money.
Cost of money: You could borrow the money. Then the cost of the money is the interest on the loan. Or you could use the money for something else, eg buying safe government bonds. In that case, the cost is an opportunity cost. It’s what you miss out on by not investing elsewhere.
Expected payout: It’s the average profit/loss on each book. It is something you estimate based on experience.
The more books there are on the market, the lower the average profit. There must be a limit to how much of their income people are willing to spend on books. At some point, you have a lot of similar books chasing the same audience. That lowers the average. To maximize your profit, you invest in the production of more and more books, until the average return on each book is equal to the cost of money.
Yes. That’s economy and investment how we usually do it today. The conclusion of that is, the “manufacturers” sell their product at the end of the day. I think in the realm of what we’re discussing, it means an AI company is then the client of the book authors. And they pay for the books, or more the content within. That’s the traditional model and doesn’t make sense unless it results in some product being sold.
You suggested that the government should produce textbooks to prevent abuse. Would that also be your solution here? Would that be preferable to the current arrangement?
Now that’s a really interesting question. Some intelligent people have proposed similar things, economy being controlled by the government instead of the free market. And we’ve tried it. Turns out it’s tricky to get it right. When they tried applying it to the entire economy, it often resulted in lots of corruption, an underperforming economy, up to outrageous things like famine and starvation in the population. Though I’m making it sound simpler than it is. Lots of different factors were involved with that.
And then sometimes we get it somewhat right. For example education is done by the government. Public infrastructure like roads, trains… And the government already produces books and TV. One example is public broadcasting like the BBC or ARD/ZDF here. I think what they produce is far superior than news in the USA. On the downside it’s a very bloated organization and they waste lots and lots of money doing it.
So… My answer to your question is: yes and no. Yes, government should produce books and other content. Like local news from my region, which is not a profitable business so the private companies regularly fail due to that. And education would be another topic. It’d be great if education were accessible to everyone, at no cost. Maybe some other things.
And no, I don’t think government should produce all books and content. That’d be kind of a monopoly on information. It’s hard to choose which book should be written and which discarded. Which wannabe autor to put on the payroll… We’d need a lot of trust and faith in the government, which we don’t have. And it’s likely going to fail because of a multitude of reasons. I’d say it’s somewhat a nice idea. But I give it zero chance to work as intended in reality.
I think in the realm of what we’re discussing, it means an AI company is then the client of the book authors
Ahh. But they are not. That’s what we’re discussing.
Let me make this clear: All intellectual property is arbitrary. I fear many copyright people have convinced themselves otherwise.
The government could grant the exclusive right to sell coffee in an area. That was done at one point. It could give the exclusive right to make shoes to some corporation. That was normal before the time of the French Revolution. The German constitution explicitly protects the right to chose one’s profession. The origin of this lies in such feudal practices.
The US Constitution limits copyright because the founders were quite aware of how these feudal privileges were abused. European copyright descends from agreements between mostly monarchical empires. Rent-seeking was/is an intended feature, which is why Europeans are so easily defrauded by the copyright industry.
When you photograph an image, you have to get permission. Makes sense. When that image is in the background of a video, you may have to get permission. Makes less sense. You rarely have to get permission from makeup artists, hairdressers, and clothes designers. Why not, actually? Isn’t that “theft” on a grand scale?
Historically, it makes sense. Originally, copyright was for printing. The only images you could print were engravings. It would have been hard to justify that the tailors, maids, or butlers should get a cut. And also, they were not a demographic that could expect to be favored with an economic rent from the elites.
And today? There are many photos that derive more value from the clothes and general appearance of the model than from anything else. And yet, the photographer owns the copyright and only needs to get permission from the model. How should that work?
By the by. Painters and some intellectuals raged against photography in much the same way that they rage against AI now. There is an essay by Charles Baudelaire that illustrates this nicely.
Ahh. But they are not. That’s what we’re discussing. Let me make this clear: All intellectual property is arbitrary.
I feel we’ve ran into the exact same issue as before. Now we’re talking property. But we were just talking about investment and we’ve just established those two are distinct and not the same. It’s a bit confusing. And I agree, that resulting granted monopoly and rent-seeking is an intended feature, and not contributing to society. But my previous comment was addressing the aspect of the author’s investment and ROI, not the resulting property from that. And that’s not arbitrary at all. The author sat at his desk for 6 months specifically. Sure the resulting product is arbitrary when selling it for money, but that wasn’t what we were talking about.
which is why Europeans are so easily defrauded by the copyright industry
I don’t think we’re easily defrauded by the copyright industry. As I said, school-books seem like 10x cheaper here. Medication with pharma IP in it is mostly cheaper here, I have my library card for like 30€ a year?! And other than that we use the same Spotify and Netflix subscriptions for a similar price. There’s no substantial difference with that. I don’t see myself in a less favourable position than an US citizen. We also have access to information here, good books, podcasts, journalism, we have culture, concerts… And I don’t think any of that is better or cheaper or more accessible in the US. Correct me if I’m wrong…
photograph
Yeah, some photography rules are absurd. I think it’s completely mental that people do copyright infringement when they take a picture of a sculpture. Seems US Fair Use sometimes has weird quirks. We also have stupid rules for pictures in Germany.
[…] feudal practices
Considering feudalism… I’d like to re-define that since wo don’t have lords and a king for quite some time now. Today’s land holders on the internet are companies like Meta, Google etc. They own the platforms we use on a daily basis. They make the rules, shape the place and lease chunks to us peasants as a service. We even let them shape society. For all intents and purposes, they’re the feudal lords of today. And that’s kind of the reason for my rejection here and why I said early on, all these AI companies are big multi-billion dollar corporations with motivations far from benefit to society. I believe concepts like Fair Use might have been invented as a means to combat feudalism. But looks to me like the situation is now changing and it’s more and more used to the opposite effect by the feudal lords themselves to now contribute to their posessions, wealth and dominance.
I’ll grant you the copyright industry is a worthy enemy, since they’re villains, too. The copyright business model isn’t healthy or beneficial to society overall. We’ve established that. But I really think of feudalism and a defacto-monopoly when I think of Google and Meta and OpenAI/Microsoft. And I’d really like to avoid making more concessions to my feudal lords.
Hmm. It looks like we are back to narratives again. Systematic analysis does not seem to come easy to you.
Now we’re talking property. But we were just talking about investment and we’ve just established those two are distinct and not the same.
“Investment” and “rent-seeking” are concepts in economics. Like, say, “function” or “variable” are concepts in programming.
“Property” is a legal institution. It relates to “investment” a bit like a machine code instruction relates to programming. They are, sort of, the underlying facts on which higher concepts rest.
And that’s not arbitrary at all. The author sat at his desk for 6 months specifically. Sure the resulting product is arbitrary when selling it for money, but that wasn’t what we were talking about.
I guess you didn’t get what I was trying to say. Let me put it like this:
If they wrote a story that takes place in the universe of a video game, then they need to get permission first. They need to ask whoever owns the rights to the video game, or else it is “theft”.
Conversely, if the story is original, and anyone wants to make a video game in that universe, then they need the author’s permission.
This remains so until 70 years after the death of the creator of the video game/story. At least, it is 70 years now. It may be made longer again at any time.
That is arbitrary, no?
Today’s land holders on the internet are companies like Meta, Google etc.
Not just them, but yes. How do you think they manage that?
And that’s kind of the reason for my rejection here
That seems pretty vibes-based. What do you rationally expect the outcome of your favored policies to be?
Yes. That’s arbitrary. But we’re conflating several very different things here. There is investment in form of labour. And I’m pretty sure we have to agree that in general, labour needs to be compensated in a capitalist economy. Then there is copyright. And this is intellectual property, which is yet another concept. All of this goes into a book, but they’re all very different things. I think IP is the most abstract one (it protects concepts) and kind of moot. I’d be more lax with IP and try to allow everyone to draw a Mickey Mouse, program a Final Fantasy game or write a new Harry Potter book. Patents are a similar thing. Though we have them for a reason.
That’s why I say I’m with you with the copyright and the intellectual property. But there’s also work going into a book and we’re always brushing over that as if it weren’t a thing.
How do you think they manage that
It’s many factors. Timing, aggressive acquisition strategies, ecosystem building, network effects, then ecosystem lock-in, data harvesting, dominating standards, but also providing genuinely useful services. Economy of scale, massive capital… And I probably forgot dozens of factors, some legitimate, some exploitative.
That seems pretty vibes-based. What do you rationally expect the outcome of your favored policies to be?
A more level playing field for new players and institutions apart from mega-corporations
More transparency, since this is a disruptive technology with impact on society
Expanding on transparency: Mandating transparency in cases like: Why was my loan declined? Why is my insurance now 4x the cost? And is the picture/text on the internet misinformation and fake or real?
More public research and access to AI. AI shouldn’t be just a for-profit service shaped by the tech bros
Regulation of Black Mirror episode content, like social scoring, total surveillance and mass control, fraud and big-scale manipulation of people, discrimination… And oversight and mandatory standards for dangerous tech, like systems used in healthcare or the arms industry.
Handle copyright in a way that applies universally. It’s unfair and deeply undemocratic to allow Mark Zuckerberg to pirate books because he’s rich and has an AI company, while I and other businesses can go to jail for the exact same thing.
Less ruthless business practices like deliberately abusive data scraping.
Clarify edge-cases like whether it’s okay to impersonate Scarlett Johannsson or David Attenborough. Or generate pornography of Emma Watson.
Incentives to develop open-weights models (ideally more than that) and to contribute to society and progress.
Sorry, misunderstanding. I wasn’t asking what you hope to happen.
You have ideas on how copyright should work wrt AI training. Make these ideas explicit, and then try to systematically analyze what the economic effects are.
Law can be a little bit like programming. A law has certain conditions. If these conditions are met, then certain legal effects follow.
If certain conditions are met, then someone has the exclusive copyright. If this copyright is violated, then damages must be paid. And of course, there are more rules to determine if copyright was violated or how those damages should be determined.
So under what conditions does AI training violate copyright? What would the legal consequence be? Then, what would that mean for the economic system on the whole?
That’s a tough question. Copyright is showing its age and barely applies in the digital world. Even before AI we had a lot of edge cases and court cases over like a decade to find out how copyright applies to a digital concept. I don’t think there is an easy way to retrofit something. At least I can’t come up with a good idea. And the general proposal seems to be all or nothing.
What I think doesn’t work is saying every normal citizen needs to buy books and Zuckerberg gets to pirate books. In a democracy law has to apply to everyone. And his use-case doesn’t matter here. I can also claim I pirated the 10TB of TV shows and movies for transformative or legitimate use. It’s still piracy. And other law works the same way. If I steal chocolate in the supermarket, that’s also theft no matter what I was planning to do with it. So that’s out.
And then we’re left with how economy is supposed to work as of today. An AI company needs supplies to manufacture their product, they buy those supplies on the market… In this case that’s going to be licensing content. Though, that’s going to be hard. A billion dollar company with a service used by millions of people should pay more than a single researcher doing it for 5 people. And implementing that would be impossibly complex. One possible way would be to introduce a collecting society to handle the money and maths. But they’re not ideal either.
So it’s more or less down to allowing AI companies to use content with some kind of default license. They can take all the public information as they wish. Again, they can not steal in the process. They’ll buy one copy of a Terry Pratchett novel at the same price everyone needs to pay.
And to compensate for them not having to contract with the authors an buy special licenses, they need to offer transparency. Tell the authors and everyone what went into the models and if their content is amongst that. And if they scraped my personal data, I need a way to get that deleted from the dataset.
I’d also add an optional opt-out mechanism to appease to the people who hate AI. They can add some machine-readable notice, or file a complaint and their content will be discarded.
And since just taking and not contributing back isn’t healthy to society, I’d add something about “composite” works. If something like an AI model is just pieced together by other people’s content, that doesn’t deserve copyright in my opinion. So all generations are automatically public domain and maybe the models as well.
And we need a definition of AI and transformative. Once we get capable models with a ability to recite an entire novel word by word, that’s going to run into copyright again. So yeah.
And intellectual property has to be softened. A generative AI model necessary “contains” a lot of IP, has knowledge about it and can reproduce it. And we need to be alright with that. And in case someone wants to outlaw impersonation and celebrity deepfakes, there needs to be more than a blurry line.
But all of this is more patching copyright and we’re going to run into all kinds of issues with that. I think ideally we come up with a grand idea and overhaul the entire thing so it applies to the 21st century.
I don’t think my opinion as some random dude matters here. I could uphold arbitrary stupid believes. But this is kind of a factual question. So whether I personally, as one person, am fine with something is of no concern here. The question is, how do we arrive at a consistent economy model for immaterial goods…
And I think I wrote like 5 times now that I’m NOT fine with that. I said I view it as a (necessary) evil. It is evil in the sense of bad, I’m not fine with it, it comes with severe issues, we should do better than that. However “is” and “should” are two seperate things. We happen to live on a world that came up with copyright. It exists. We made a pact with the devil to address one thing. And I’m merely acknowledging that. Since it does exist, I need to deal with it. That’s not agreement from my side. Copyright serves one legitimate purpose. It applies our capitalist economy to immaterial goods. It’s supposed to allow individuals and companies to create, and trade with more than just cocoa beans. But it’s complicated and we might have come up with a stupid way to do it. And a way that simultaneously has lots of negative side-effects.
And now what? That is the question. Do we abolish it? Do we replace it with something else that handles the one legitimate purpose a better way? Do we retrofit it and try to “patch” it? Do we do that just for AI? Or for more than just one use-case?
And I think I make a point about how return on investment and an economic rent are two distinct things. Yet they’re in practice falsely(!) mushed together, which again is bad… Or am I mistaken and I can pay an artist for their investment but not pay a rent? I don’t think there is a good way to do it with the current model. That means I get to treat both as the same. You seem to be under the impression I like it. But I don’t. It’s just that I have to abide by law and that currently mandates me to do it.
I see. I think this is the big one:
A return on investment is not the same as an economic rent.
Let’s go back to the farmer example. You agree that a monopoly on the food supply is a bad thing. It can and will be abused.
Sidenote: You suggested that the government should produce textbooks to prevent abuse. Would that also be your solution here? Would that be preferable to the current arrangement?
Now, let’s look at the situation of a farmer more closely. A farmer has to do a lot of work before they can harvest. They also need stuff like seeds, fertilizer, pesticides, fuel, machinery, spare parts and maintenance, and so on.
In the old times, one held back part of a grain harvest as seed grain for next year. That is an investment in the economics sense. You don’t consume everything, but keep it so that you have more in the future. The finance meaning is subtly different but never mind.
Farmers gets a return on investment. They invest money and labor so that there is a harvest in the future. They could sell the equipment they already own to have more spending money now.
A ROI is part of a farmers’ income but is not economic rent.
Back to authors. An established author will get an advance before they write the next book. That’s investment by the publisher. If they don’t get an advance, then the author is making the investment, but let’s ignore that for simplicity. Investments are always risky. In this case, some books don’t sell well and don’t make back the money.
As a publisher, how much money would you invest in future books to maximize your profit? It depends on the expected payout and the cost of money.
Cost of money: You could borrow the money. Then the cost of the money is the interest on the loan. Or you could use the money for something else, eg buying safe government bonds. In that case, the cost is an opportunity cost. It’s what you miss out on by not investing elsewhere.
Expected payout: It’s the average profit/loss on each book. It is something you estimate based on experience.
The more books there are on the market, the lower the average profit. There must be a limit to how much of their income people are willing to spend on books. At some point, you have a lot of similar books chasing the same audience. That lowers the average. To maximize your profit, you invest in the production of more and more books, until the average return on each book is equal to the cost of money.
I’ll leave it at that for now.
Yes. That’s economy and investment how we usually do it today. The conclusion of that is, the “manufacturers” sell their product at the end of the day. I think in the realm of what we’re discussing, it means an AI company is then the client of the book authors. And they pay for the books, or more the content within. That’s the traditional model and doesn’t make sense unless it results in some product being sold.
Now that’s a really interesting question. Some intelligent people have proposed similar things, economy being controlled by the government instead of the free market. And we’ve tried it. Turns out it’s tricky to get it right. When they tried applying it to the entire economy, it often resulted in lots of corruption, an underperforming economy, up to outrageous things like famine and starvation in the population. Though I’m making it sound simpler than it is. Lots of different factors were involved with that.
And then sometimes we get it somewhat right. For example education is done by the government. Public infrastructure like roads, trains… And the government already produces books and TV. One example is public broadcasting like the BBC or ARD/ZDF here. I think what they produce is far superior than news in the USA. On the downside it’s a very bloated organization and they waste lots and lots of money doing it.
So… My answer to your question is: yes and no. Yes, government should produce books and other content. Like local news from my region, which is not a profitable business so the private companies regularly fail due to that. And education would be another topic. It’d be great if education were accessible to everyone, at no cost. Maybe some other things.
And no, I don’t think government should produce all books and content. That’d be kind of a monopoly on information. It’s hard to choose which book should be written and which discarded. Which wannabe autor to put on the payroll… We’d need a lot of trust and faith in the government, which we don’t have. And it’s likely going to fail because of a multitude of reasons. I’d say it’s somewhat a nice idea. But I give it zero chance to work as intended in reality.
Ahh. But they are not. That’s what we’re discussing.
Let me make this clear: All intellectual property is arbitrary. I fear many copyright people have convinced themselves otherwise.
The government could grant the exclusive right to sell coffee in an area. That was done at one point. It could give the exclusive right to make shoes to some corporation. That was normal before the time of the French Revolution. The German constitution explicitly protects the right to chose one’s profession. The origin of this lies in such feudal practices.
The US Constitution limits copyright because the founders were quite aware of how these feudal privileges were abused. European copyright descends from agreements between mostly monarchical empires. Rent-seeking was/is an intended feature, which is why Europeans are so easily defrauded by the copyright industry.
When you photograph an image, you have to get permission. Makes sense. When that image is in the background of a video, you may have to get permission. Makes less sense. You rarely have to get permission from makeup artists, hairdressers, and clothes designers. Why not, actually? Isn’t that “theft” on a grand scale?
Historically, it makes sense. Originally, copyright was for printing. The only images you could print were engravings. It would have been hard to justify that the tailors, maids, or butlers should get a cut. And also, they were not a demographic that could expect to be favored with an economic rent from the elites.
And today? There are many photos that derive more value from the clothes and general appearance of the model than from anything else. And yet, the photographer owns the copyright and only needs to get permission from the model. How should that work?
By the by. Painters and some intellectuals raged against photography in much the same way that they rage against AI now. There is an essay by Charles Baudelaire that illustrates this nicely.
I feel we’ve ran into the exact same issue as before. Now we’re talking property. But we were just talking about investment and we’ve just established those two are distinct and not the same. It’s a bit confusing. And I agree, that resulting granted monopoly and rent-seeking is an intended feature, and not contributing to society. But my previous comment was addressing the aspect of the author’s investment and ROI, not the resulting property from that. And that’s not arbitrary at all. The author sat at his desk for 6 months specifically. Sure the resulting product is arbitrary when selling it for money, but that wasn’t what we were talking about.
I don’t think we’re easily defrauded by the copyright industry. As I said, school-books seem like 10x cheaper here. Medication with pharma IP in it is mostly cheaper here, I have my library card for like 30€ a year?! And other than that we use the same Spotify and Netflix subscriptions for a similar price. There’s no substantial difference with that. I don’t see myself in a less favourable position than an US citizen. We also have access to information here, good books, podcasts, journalism, we have culture, concerts… And I don’t think any of that is better or cheaper or more accessible in the US. Correct me if I’m wrong…
Yeah, some photography rules are absurd. I think it’s completely mental that people do copyright infringement when they take a picture of a sculpture. Seems US Fair Use sometimes has weird quirks. We also have stupid rules for pictures in Germany.
Considering feudalism… I’d like to re-define that since wo don’t have lords and a king for quite some time now. Today’s land holders on the internet are companies like Meta, Google etc. They own the platforms we use on a daily basis. They make the rules, shape the place and lease chunks to us peasants as a service. We even let them shape society. For all intents and purposes, they’re the feudal lords of today. And that’s kind of the reason for my rejection here and why I said early on, all these AI companies are big multi-billion dollar corporations with motivations far from benefit to society. I believe concepts like Fair Use might have been invented as a means to combat feudalism. But looks to me like the situation is now changing and it’s more and more used to the opposite effect by the feudal lords themselves to now contribute to their posessions, wealth and dominance.
I’ll grant you the copyright industry is a worthy enemy, since they’re villains, too. The copyright business model isn’t healthy or beneficial to society overall. We’ve established that. But I really think of feudalism and a defacto-monopoly when I think of Google and Meta and OpenAI/Microsoft. And I’d really like to avoid making more concessions to my feudal lords.
Hmm. It looks like we are back to narratives again. Systematic analysis does not seem to come easy to you.
“Investment” and “rent-seeking” are concepts in economics. Like, say, “function” or “variable” are concepts in programming.
“Property” is a legal institution. It relates to “investment” a bit like a machine code instruction relates to programming. They are, sort of, the underlying facts on which higher concepts rest.
I guess you didn’t get what I was trying to say. Let me put it like this:
If they wrote a story that takes place in the universe of a video game, then they need to get permission first. They need to ask whoever owns the rights to the video game, or else it is “theft”.
Conversely, if the story is original, and anyone wants to make a video game in that universe, then they need the author’s permission.
This remains so until 70 years after the death of the creator of the video game/story. At least, it is 70 years now. It may be made longer again at any time.
That is arbitrary, no?
Not just them, but yes. How do you think they manage that?
That seems pretty vibes-based. What do you rationally expect the outcome of your favored policies to be?
Yes. That’s arbitrary. But we’re conflating several very different things here. There is investment in form of labour. And I’m pretty sure we have to agree that in general, labour needs to be compensated in a capitalist economy. Then there is copyright. And this is intellectual property, which is yet another concept. All of this goes into a book, but they’re all very different things. I think IP is the most abstract one (it protects concepts) and kind of moot. I’d be more lax with IP and try to allow everyone to draw a Mickey Mouse, program a Final Fantasy game or write a new Harry Potter book. Patents are a similar thing. Though we have them for a reason.
That’s why I say I’m with you with the copyright and the intellectual property. But there’s also work going into a book and we’re always brushing over that as if it weren’t a thing.
It’s many factors. Timing, aggressive acquisition strategies, ecosystem building, network effects, then ecosystem lock-in, data harvesting, dominating standards, but also providing genuinely useful services. Economy of scale, massive capital… And I probably forgot dozens of factors, some legitimate, some exploitative.
Sorry, misunderstanding. I wasn’t asking what you hope to happen.
You have ideas on how copyright should work wrt AI training. Make these ideas explicit, and then try to systematically analyze what the economic effects are.
Law can be a little bit like programming. A law has certain conditions. If these conditions are met, then certain legal effects follow.
If certain conditions are met, then someone has the exclusive copyright. If this copyright is violated, then damages must be paid. And of course, there are more rules to determine if copyright was violated or how those damages should be determined.
So under what conditions does AI training violate copyright? What would the legal consequence be? Then, what would that mean for the economic system on the whole?
That’s a tough question. Copyright is showing its age and barely applies in the digital world. Even before AI we had a lot of edge cases and court cases over like a decade to find out how copyright applies to a digital concept. I don’t think there is an easy way to retrofit something. At least I can’t come up with a good idea. And the general proposal seems to be all or nothing.
What I think doesn’t work is saying every normal citizen needs to buy books and Zuckerberg gets to pirate books. In a democracy law has to apply to everyone. And his use-case doesn’t matter here. I can also claim I pirated the 10TB of TV shows and movies for transformative or legitimate use. It’s still piracy. And other law works the same way. If I steal chocolate in the supermarket, that’s also theft no matter what I was planning to do with it. So that’s out.
And then we’re left with how economy is supposed to work as of today. An AI company needs supplies to manufacture their product, they buy those supplies on the market… In this case that’s going to be licensing content. Though, that’s going to be hard. A billion dollar company with a service used by millions of people should pay more than a single researcher doing it for 5 people. And implementing that would be impossibly complex. One possible way would be to introduce a collecting society to handle the money and maths. But they’re not ideal either.
So it’s more or less down to allowing AI companies to use content with some kind of default license. They can take all the public information as they wish. Again, they can not steal in the process. They’ll buy one copy of a Terry Pratchett novel at the same price everyone needs to pay.
And to compensate for them not having to contract with the authors an buy special licenses, they need to offer transparency. Tell the authors and everyone what went into the models and if their content is amongst that. And if they scraped my personal data, I need a way to get that deleted from the dataset.
I’d also add an optional opt-out mechanism to appease to the people who hate AI. They can add some machine-readable notice, or file a complaint and their content will be discarded.
And since just taking and not contributing back isn’t healthy to society, I’d add something about “composite” works. If something like an AI model is just pieced together by other people’s content, that doesn’t deserve copyright in my opinion. So all generations are automatically public domain and maybe the models as well.
And we need a definition of AI and transformative. Once we get capable models with a ability to recite an entire novel word by word, that’s going to run into copyright again. So yeah.
And intellectual property has to be softened. A generative AI model necessary “contains” a lot of IP, has knowledge about it and can reproduce it. And we need to be alright with that. And in case someone wants to outlaw impersonation and celebrity deepfakes, there needs to be more than a blurry line.
But all of this is more patching copyright and we’re going to run into all kinds of issues with that. I think ideally we come up with a grand idea and overhaul the entire thing so it applies to the 21st century.