Panel Discussion AI and the Creative Economy

Yes. Yes. three crucial elements. I’ll start with Nicholas Granatino, who’s on the business side, the chairman of Tara Gaming. Second is Kenichiro Natsume, who is the assistant director general at WIPO on the policy side. And we have Anna Tumadote, who is the chief executive officer of Creative Commons. Since we have a half an hour, I don’t want to waste time. I have this urge to do a lot of context setting, but I will refrain. And I will jump into the first question for each of the panelists, and then we can do it more conversational. First question to each of the panelists. Anna, can we start with you? Does AI strengthen or weaken cultural diversity in the global creative output?

It’s a good question, and I think we can just do our context setting along the way here with this. So I think this is one of those wonderful questions where the answer is going to be, it depends. It depends. Is the model open source, and we can all interrogate what’s in it and build upon it and improve it? Or is it closed source? Is the model got good governance frameworks attached to it so we can understand some of the intentions behind it or is it all very opaque? So I think ultimately it’s going to just come down to what sorts of values and design principles we’re able to instill as we build this new ecosystem.

I think currently we are at risk for a weakening.

Thank you very much. My answer is not a binary answer. Because if we think about the intellectual property aspect, namely copyright, then AI is okay. AI can be used to enhance the copyright table or artworks. At the same time, that could be also kind of a threat because people can create something or generate something by using artificial intelligence. and in terms of the legal perspective or international perspective we do have a copyright system which in my personal opinion can still cope with the artificial intelligence because I think that artificial intelligence at this stage is one of the cutting edge technologies which is magnificent but one of the cutting technologies or disruptive technologies we have been experienced in the history of our human life.

Thank you.

Yeah. Namaste. I think I will actually give an answer which is very positive on the creativity side and that’s what we saw at Atara Gaming as an opportunity is the fact that from the web to social graph to now AI I think we always underestimate and don’t talk enough about the data and the content and the case of AI is what is called the latency space which includes the training data set. which is used for these models to respond to prompts. And if we look at the case of India, which has been mainly an oral tradition, and we look also at how much of their content have been digitized to the level of Hollywood in movies or AAA game in gaming, it is not represented.

And that has huge implication in terms of these models and whether or not they enhance creativity or whether or not they need creativity. And my answer to that is they need creativity. They need actually the wonderful epic and story and living traditions that are in the ETS, which are these Indian epics, to be represented in their data set. And at the moment, they’re not. And so the bigger question for me is how do we make sure that India, with its phenomenal history and culture and 20 % of the world population, punches at its weight? in the training data sets of these AI models. And the first work that needs to be done is a creative work. It’s not an AI work.

You know, Nicholas, when I sent these questions, the kick I got was Nicholas pushed back happily and he gave me three questions which I told him that one of them really fired my imagination. I want to ask him about this in terms of this aspect and this whole AI IP sort of dispute in terms of the US and Europe are currently facing massive uncertainty in terms of AI training and copyright assets and permission. And that’s either, depending on who you are, it’s either stalling, you know, development, etc. You’ve said, Nicholas, that this paralysis is India’s biggest opportunity. How does India’s rich public domain heritage, the itihas, as you called it, give us a strategic edge right now?

Yeah, so the ETS, actually, maybe most of you know one part of the Mahabharata, which is one of the ETS. The second one is the Ramayana. The one I’m referring to is the Gita. And so maybe most of you are more familiar with the Gita, which is actually just a short part of the Mahabharata. These are epics that are thousands of years old. What’s phenomenal about them is they are in the public domain, but they are still living tradition. Parents and grandparents still tell their children about Lord Ram, about Ravan, about Hanuman. Prime Minister Modi speaks about it all the time as well. And they are in the public domain. And the reason, and nobody has actually done any work with them and created any IP on top of that, the quality we’re looking to do at Tara Gaming.

And so there is basically a data set. And what’s really exciting about Sarvam AI is that they’re actually going to help. to actually digitize with their OCR model, with their voice model, all this culture which is in the public domain and which can enter through SARVAM and through the works that we’re doing, actually the data set to train. And that creates an opportunity, I think, which is quite unique. And the second layer is the fact that you have 20 % of the world population here in India to talk about it and create rich data sets about that. And the interesting point is that on this public domain content and body of work, you have this opportunity to create IP.

And Ken, the question is for you. Is the global IP framework prepared for large -scale AI -generated content today?

Thank you very much for the very interesting but big question. It’s not very easy to answer in a short time, but let me try. Because I… Artificial intelligence, particularly in the area of the creative industry, of course, it’s changing the work of the artists and creators. and I’ve been discussing with let’s say in India for example taking advantage of my visit to Delhi meeting with different stakeholders publishers music industry or tech industries and the views are very much different it’s no secret and you know that the views are very much different that’s the reality in front of us and even among the same segment for example publishers there are different views and for example artists, one of my friend artists, she’s using AI to create digital art at the same time the other artists are kind of feeling threatened by the generation of the artwork or the output generated by so called generative AI so it’s not very straightforward and the reality in front of us is that it’s not There are very much different views and it’s not so easy to find a common denominator because the views are so different.

It’s not something like zero and one and let’s see 0 .5. It’s not like a mathematic. So if we see about the intellectual property system, at this stage our view is the following because that we, WIPO, as a UN international organization sometimes are asked that, hey WIPO, why don’t you think about making some rules or regulations, international treaty of AI and intellectual property. Not sounds bad, but how long does it take? We have 194 member states. Our principle is consensus. Okay. So 194, meaning 194 member states, including of course India, should agree upon one common thing. and just to be frank with you, don’t quote me, it’s a long journey. Sounds relatively easy. Yes, yes. So our approach is more pragmatic.

Okay, we can think of something, legal framework, but let’s put it aside because it’s not, the international scene is not matured enough. It’s not ready enough. So our idea is let’s think about something more practical, more technological, to see if there is any technological solution possible so that both creators’ side and industry side, tech industry side, can live with. I would say can live with. They don’t have to be necessarily happy very much. They have to be unhappy to some extent equally. They have to compromise each other. But it’s not an international legal negotiation. It’s a collaboration or cooperation explore to find some technological infrastructure where people can live with us so that creators can be benefited or remunerated as well as tech industries can utilize those products.

Thank you.

Anna, you know one of the things Ken said is an artist and use AI to create artworks and you have artists today and if I look at the music industry film industry you have composers who use AI to generate production music and a lot of musical works maybe not declared or otherwise but it’s a huge industry but at the same time the creator voice often complains about the lack of consent in AI training whereas most of them work on AI models which have trained on a worldwide corpus of data and content. Do you see a sort of ethical or rational inconsistency here in this sort of kind of use and objection or what do you think is the correct position?

Is there an ethical inconsistency? Yes, yes is the answer. It’s funny that you get the, hey, WIPO, can you fix this? Because Creative Commons gets that from time to time, too. In fact, in the early days of generative AI, we were getting a, hey, Creative Commons, can you fix AI for us? We’re like, okay, which direction are we going to fix it in? I’m just kidding. We never suggested that we would actually fix it. But it actually comes down to this question that you were asking. So, you know, here we have the world’s creativity that has been scraped, crawled, trolled, whatever, you know, however you want to describe it. And we built these massive foundation models.

And the, you know, the sort of relative weight of every individual work in there is infinitesimal. It’s tiny. It’s tiny. The bigger concern is what does use of these technologies do to the creative industries, right? It’s more a fear of, like, replacement. It’s a labor issue. It actually feels increasingly less. Like a copyright issue when we’re talking about some of these considerations. But then there’s that layer on top in the sort of inference level where. you know, you’re querying, like, tell me, you know, tell me a story about this, you know, tell me, tell me, you know, about a certain concept, or whatever the case may be, and we’re not seeing where that information is coming from, right?

So, we’re sort of divorced from where the origin of the creativity, or the knowledge, or whatever it was, came from, and that, I think, is going to be a longer -term problem for these tools, because you’re not really going to trust in how they work, and you see it show up similarly with the artistry piece, right? So, we have artists who have always embraced the free culture movement, you know, give things over to the public domain, or give things over with a creative commons license, and are enthusiastically experimenting with these technologies. We have artists, too, who have, like, vast bodies of work, where they’re building their own models, and so they’re just enhancing their own craft.

So, interesting examples to look at would be Holly Herndon and Imogen Heap, who’ve, like, really been on the forefront of this, um, but at the same time, it was to your point, you know, there are artists who are playing with this, but are like, no, no, no, anything that I create, that’s mine, but I’m going to use all the world’s creativity here freely, and we have to find some kind of a middle ground in this, because Ultimately, all knowledge builds on prior knowledge. All creativity builds on prior creativity. The richness of the public domain, like Nicholas was talking about, you can walk into any museum and be inspired. You don’t have to go, I saw this work and that work and that work and that work and that work, and now I’ve sketched this drawing.

But there is something with the technological layer where if you are fusing together different things or asking for certain styles or certain inspirations or so on, there really should be some sort of form of credit given.

Please, Nicholas.

Yeah, I mean, I think everybody has celebrated, including the Nobel Committee, the work of Demis Hassabis at DeepMind on protein folding. But the reality is there have been actually scientists which have been for 50 years putting their crystal structure of proteins into a database called the Protein Database. I think it would have been nice of the Nobel Committee to also include the protein data bank as a recipient of that Nobel Prize. And actually, the data has always been put under the carpet. I mean, a lot of the big tech is saying content is free, what people pay for is search, tech, AI, whatever it is, the link, the graph, the link graph, the social graph, and now the AI model.

And so, the question they want to pose is, do you want AI as a society to have the best data? And the answer is probably yes on the condition that you are open, but if you are going to make money at the gateway of the chat or whatever is your application, and you’re going to assume that everybody works for you, I don’t think society wants that. President Macron yesterday talked about it’s not about regulation, it’s about civilization. And I think it’s, as a civilization, what do we want as a future? How much do we want to rework the work of these protein crystallographers, of these creatives, etc.? That’s the real question.

Please, Ken.

Just one quick note. Anna’s comment was very touching to me because you mentioned it. we human beings refer to or learn from the other person’s creative work, which is true. I’m a Japanese, and the Japanese term of learn originally meant mimic or copy. So the learning is starting from to look at some other person’s work and try to imitate, and based on that we develop our own flavor or texture. And this has been done by human beings for ages. The big difference is that if it’s done by human, it takes time. But if it’s done by computer, it takes very, very limited time. And that’s a big difference. So what it’s doing is essentially more or less the same, but the speed is completely different.

And maybe we have to draw a line. This is what exactly Anaba was saying. And the answer, where do we have to draw a line? Is a difficult question.

You know, it’s so funny you say that because even in the open movement and in the open knowledge communities, we’ve had problems for years, but they’ve been problems on the margins, right? They’ve been the sort of 5 % problem, and we’ve thought to ourselves, you know, this is 95 % good, and so that’s good enough. But AI comes along, and the scale is so massive that now you have to grapple with this. So, for instance, like, what if you’ve shared your work freely, and then it’s used for nefarious purposes? Nobody wants that. Like, that’s the sort of, like, another ethical conundrum that you face. But copyright is not built to handle that. Like, society ideally would find ways around that.

Maybe there’s normative frameworks we need to introduce. Maybe we need to think about sort of different legal or technical solutions to this because the scale is just so extreme.

So, Ana, is the open movement something that can withstand this, the AI onslaught or the spread of AI? Is it structured to do that, or would it have the same challenges that, let me call it the proprietary copyright model as the traditional… The traditional copyright model.

I think there’s a transformation that’s going to have to happen because I think what we’re actively seeing is people pulling back from sharing their works because, you know, for whatever, you know, if they have no consent or, you know, consent mechanism or no agency over how it’s used, they’re going to probably do the only thing that they can in that situation and either take it back or put it behind a wall or try to make you pay for it or, you know, any of these other levers that they can pull. And we are actively seeing the shrinking of the commons already. Like, this is a really bad outcome. And here’s the real kicker when we’re talking about the sort of humans have been doing this for a long time.

Like, human -to -human collaboration relies on copyright clarity, on the CC licenses, on this ability that, like, if I write something, you know what you can do with it because it’s under this license and so on and so forth. But now we’re seeing creators say, no, no, I’m going to go more restrictive. And that breaks the human collaboration element. So there’s all these downstream negative consequences from this. So I think we can withstand it, but I think collectively we have to reckon with the fact that there is a problem. and the scale is so magnificent that we can’t just stick our fingers in our ears and say, you know, ultimately this is in the public interest.

It’s not going to be that way because if nobody shares, then there’s nothing left for us.

You know, Nicholas, to your point on the opportunity for India and the use of ethos and public domain, but, you know, being the opportunity for India to create that IP layer, isn’t there also the danger that we might slip into this cultural gatekeeping by a handful of AI platforms? Is that then an overwhelming danger as well?

Yeah, absolutely. I mean, I think, you know, Europe and Mistral are pushing a lot for, you know, open source, and China is pushing a lot also in the open source community. But that’s just a layer above of what I think we’re talking about here. today. There’s something I call the Captain America hegemony, which is basically we all grew up in France, in India, anywhere with Captain America as being this kind of powerful with all the weapons, all the defense mechanism, etc. And if you actually have open source, you just have a gateway which is free to steal a corpus of creativity, which is in this Captain America hegemony. And I think what we want is we have to remember that AI is just on top of something that has been done, and so the creative process is not really part of that.

It’s above language, it’s above image, etc. And we know that everybody agrees in the AI community there’s actually two, three things that need to happen to reach AGI. And those two, three things probably overlap with a lot of the creativity that makes us unique and allow us to make this corpus that AI will continue to be trained of. So I’m quite optimistic because there is things that is pre – art, whether it’s text, whether it’s image, whether it’s video, whether it’s game, you know, that is still a creative process. Sometimes that involves hundreds of people. And that you’re not going to, it’s not that you’re going to have agents speaking to each other and creating a game.

We’re very far from that.

Ken, you know, to your point on we need to find a middle ground, is this, realistically, is this a level and as a copyright lawyer, the Berne Convention, Rome, the whole push to sort of a harmonized, largely sort of co -existent copyright model across the world, has generally worked. The one exception being the Broadcast Treaty, which I started working when I was a young associate and it’s still being discussed and I love those documents because I see how that conversation has changed. But is, in the context of IP and AI, is global harmonization realistic or is fragmentation something? you know that we’ll all have to live with

i wish i could immediately say yes however the reality is as i mentioned briefly before that having a consensus with 194 member states including this country and big other countries is not always easy so that’s why we are opting at this stage for a little bit softer approach so that the technique technological solution or technological platform or technological infrastructure could actually solve the issues where creators can be rewarded or remunerated appropriately where the tech companies can easily recognize what is opted in what is opted out what is the artwork made by generated by the artificial intelligence what is done by the human being so that they can understand what can be done and what cannot be done.

So that’s the approach we are taking place. And just for your information, we will launch the first meeting of that next month, March 17th, which is available online. So please stay tuned. Thank you.

Oh, it’s in our calendar. Yeah. Great, thank you very much. Yeah, we’ll be there because I was thinking about this sort of global standard and framework question. And one of the things that we’ve tried to do with the Creative Commons community is think about, all right, what are the things that everybody wants in this moment? What are the choices they want? And what are the sort of conditions under which they would share, right? And you can imagine it going everywhere from, to your point, like in the EU, there’s the opt -out. It’s like, no, I’m not interested in this. Very important that we maintain limitations and exceptions there, though, for research. All the way to the full yes, because maybe there’s a world where people are like, put me in, put me in and tell people who I am.

But somewhere in between there, there’s a yes if. yes if you reward me, yes if you attribute me, yes if you contribute to this project, yes if you support the open infrastructure etc. and I think we just have to get a lot more sort of creative and nuanced in that spectrum.

I sense that at least for the short -term countries are going to try to find their own you know sort of policy solutions given and hopefully and I would hope that the intention is to harmonize as much as possible because I think the implications of just the AI business across the world require harmonization and scream for harmonization and so will businesses that use those models to create more IP or IP -like content. Nicholas you sit on these boards of companies frontier lab companies like Kivita and H company if there’s one mandate to Indian investors or something like that, what would you say? I think there’s a lot of work and creators in this room you want to give to ensure that we aren’t India is just not consumer of AI 2030, what would that be?

No, I think as an investor, it’s a tremendous time. I mean, I think it feels like the internet all over again. I think there’s lots of opportunity. It’s moving very fast, you know, much faster than the internet. So it’s a bit, you know, difficult to pack the right opportunity. But I think it’s going to be a collaboration with these wonderful, you know, tools that we have and human creativity. And that is going to stay. I mean, some people say storytelling is going to be the main skill in business. That is very much a human inequality. So the future is bright.

I’m seeing this big flashing red sign which says time’s up. I don’t know, mine or the panel’s. I’m hoping it’s only the panel’s. But I’m going to do a little Don Quater thing and do one last question in terms of what single principle should guide international AI governance in the creative industries over the next decade? Ken? Ken?

That’s a big question. It says time’s up, so let me do it very briefly. I think we should put human -centered approach because still creativity comes from human beings’ activity, not from the artificial intelligence. So this is one fundamental which I should take. Thank you.

I’ll just say plus one. Just keep the humans. Keep the humans at the center.

Insightful answer as always. Thank you. Thank you to the panel. Thank you to this very engaging audience. Thank you for listening to us.