S2E4: “Beyond the Duopoly: Researching Brazilian AI”
2025-10-16 40 min
Description & Show Notes
In the fourth episode of the series Luisa Conti and Fergal Lenehan, from the Centre for Digital Interculturality Studies, talk to Emilian Franco from the University of the Bundeswehr in Munich, where he is a lecturer and a PhD scholar. Emilian talked about his research dealing with the production of Artificial Intelligence in Brazil. The conversation touched on the landscape of Brazilian AI, how cultural anthropologists research Artificial Intelligence, the metaphors of ‘devouring’ and ‘eating’, and Emilian’s research experiences in Brazil.
So, is a critical and interculturally competent AI likely, and could it perhaps be produced in Brazil?
Listen to find out!
For more information on ReDICo and the ReDICo Centre for Digital Interculturality Studies, see: www.redico.eu
To join the ReDICo Hub, go to: https://hub.redico.eu
Visit the ReDICo YouTube Channel: https://www.youtube.com/@DigInterculture
For more information on Emilian Franco’s work, see: https://hub.redico.eu/community/profile/652/emilian-franco
Transcript
Chapter 1: Introduction
Hello, everybody, and you are all very welcome to ReDICo, the podcast for digital interculturality.
My name is Luisa Conti from the ReDICo Centre for Digital Intercultural Studies at the moment based at the University of Jena. ReDICo stands for researching digital interculturality cooperatively and looks to bring the perspectives of intercultural communication and internet studies closer together. This is series two of our podcast. Series one was called “The Internet and Freedom” and is still available from all major platforms. This series is called Internet Futures and is dedicated to thinking about the internet in new ways.
With me here in this podcast is Fergal Lenehan, my colleague from the Centre of Digital Interculturality Studies.
Hi, everybody. Great to be here.
And today we have our great guest, who is Emilian Franco from the University of the Bundeswehr in Munich. So welcome, Emilian.
Yeah. Hello. Hi. Thank you for having me, Luisa and Fergal.
Chapter 2: AI and Brazil
So Emilian teaches at the university, but also he's a researcher and is completing a PhD at the Professorship for Intercultural Communication and Conflict Research, a department which is very much to be situated in the area of cultural anthropology. So Emilian, you have worked on the production and the discussions surrounding artificial intelligence in Brazil. What is the AI landscape like in Brazil? Can you sketch that out for us a bit, just to begin with?
Yeah, of course I can try to sketch like a very wide and I would say fragmented landscape of AI in Brazil, how I understand it.
So very generally I would say the field is a bit torn between global aspirations to take part in the current unfolding AI era and to be a big player in the development of AI and local urgencies or hindrances. So maybe later on I will speak more about this ambivalent positionality of the researchers at the C4AI, that's the laboratory I went to and where I did my ethnography.
But when we look at Brazil and AI more from this policy level or the legislative plans that are in place in Brazil, then we would see that Brazil has started quite early to develop a broad framework regarding AI.
For example, in 2021 already, before the European AI Act was released, Brazil had already the Estratégia Brasileira de Inteligência Artificial, so short EBIA, in place, which mirrors actually the OECD principles in the first place.
and orientates itself to, there was an expert group on AI on the EU Parliament, instructed an expert group on AI exactly in 2019. And so this EBIA Act that was in place 2021 mirrors kind of this framework, and it focuses on innovation and has no regulatory teeth, that was the main critique. And so it was backed up by the Plano Brasileiro de Inteligencia Artificial, short PBIA. And that's the plan that's in place now. It started 2024 and will go until 2028. And it's like, yeah, it's the Brazilian artificial intelligence plan. It's backed up by, I think, approximately $4 billion over the next four years. And it was titled AI for the good of all.
And it's a quite interesting document, which entails a lot of interesting aims and goals for AI across different sectors, for example, public services, industry, infrastructure, and education. And at the same time, they try to implement ethical dimensions as well.
Yeah, so you see the Brazilian state already has some regulatory things in place, some documents. There's one actual bill, the 2338 bill that kind of tries to give a legal frame. And since this bill is in place, every high risk AI system that enters Brazil needs to be approved by the state. So you can see that Brazil is kind of well prepared for this AI era. At the same time
on the micro level, there is still kind of a weak infrastructure. In fact, I would say there are kind of 11 hubs all over Brazil that try to develop AI. And that's the perspective I want to engage in or that I try to look at. So how is AI developed in Brazil? And if we narrow it down even more, then we would see that there are only two big laboratories or research facilities in Brazil. The one is the C4AI, the Center for Artificial Intelligence in São Paulo. And there's one in Recife, in the north, that also does interesting research on AI.
But Emilian, just to ask you as well, you have, obviously it's unusual enough maybe for a German researcher to work so closely with Brazilian sources and with Brazil as a context. Do you have a personal relationship to Brazil? Do you?
Oh yes, yeah I have.
So my mother is from Brazil. She migrated in the 80s to Germany and so until today I have a lot of family in Brazil. For example, my uncle and when I spend time in Brazil I always see them, I can crash the couch so I have this very deep connection to Brazil via my family and my family history.
Chapter 3: The Bi-Polar Techno-Nationalist Imaginary
So going back to the topic of your dissertation. Now you were saying you're ready to give details, then I would say let's go for it, and actually as far as I understand you are focusing on the production and discourse of artificial intelligence in Brazil and how it challenges the “bipolar techno-nationalist imaginary”. Now, so what actually is this term that you use, this “bipolar techno-nationalist imaginary”?
And how it is a challenge or how are you challenging it?
Yeah, so this bipolar frame or techno nationalist imaginary that I try to use as a starting point, I would say is the common or the dominant narrative right now.
And it sees or it puts the whole AI development in a geopolitical contest and mainly between two players, the US and China. For example, if we look at the discourse around AI, we find a book like Kai-Fu Lee's 2018 AI Superpowers in which he says quite prominently that Shenzhen is the yin to Silicon Valley's yang. So you already see here this bipolarity.
And he even means that not only economically, but very much socioculturally. So he puts the Silicon Valley on the one side and says here, AI or visions of AI are clean and lofty and are mission-driven.
And this he's implicitly relying on Barbrook and Cameron. They already wrote in the 90s a book called Californian Ideology, in which they described this whole Californian ideology as driven by a mundane optimism. The early hackers of the Silicon Valley believed in a future in which everybody will be hip and rich. I quote here the Barbrook and Cameron book from 1996.
And Lee says that in Shenzhen, we have a totally different sociocultural background. Here, development is driven by scarcity mentality, as he calls it, and here it doesn't matter where an idea came from. So there's no mission that needs to be driven to an end. It's all about appropriation, application.
And because of that, he says, China will win. And this whole narrative of US versus China finds, resurfaces then in a metaphor of the AI Cold War, as for example Jens Hälterlein put it.
And I would say that this imaginary of an AI Cold War silences other ways of talking or thinking about AI. As Taylor in a 2025 article issued, he says that right now we need a wake-up call because human rights are endangered by this geopolitical antagonism and in this kind of competitive logic fosters recklessness and hierarchical thinking and precaution, collaboration and governance are prevented or corrupted, he says.
That's what I mean with this bipolar frame. In between there's nothing. There's only this competitive frame that kind of makes collaboration, sensitivity, one could say, impossible.
It's like producing a reality through thinking of it. Is this the critique?
Yes, exactly. That's the critique because it matters how we imagine things, how we put them in place and how we try to engage with them. And it's hard to challenge this dominant narrative.
Chapter 4: An Imaginary but Also a Reality?
But is there not also sort of a reality here as well that you have sort of two different ideological approaches? You have a type of...
techno-liberalism, techno-libertarianism versus techno -authoritarianism? Does this not have, do you know what I mean, okay, there's a discourse and the discourse is self-perpetuating, sort of creates itself, but there's also something to it, is there not?
Let's see if I can follow up on that. I would say that this differentiation between this digital authoritarianism that we are witnessing right now, like on different levels and scales, of course plays into this whole climate of a kind of reckless pursuit of the superpower AI.
And there are voices that are kind of warning and issuing wake-up calls not only from academia but there was this memorandum. I think even Elon Musk said that we need to stop developing AI or Sam Altman from OpenAI as well.
So what I guess is that, and now I try to circle back to Brazil and I want to speak about a paper that I found very interesting from Figueira, Garrote and Zanatta in 2024 and they called it Cosmologies of AI, that in this climate and in this era of AI development that either leans to an ordoliberal way or into an authoritarian way, a whole worlding of AI, worlding kind of in the Haraway sense, is lost.
And they critique quite sharply that, for example, even Brazil's AI strategy, even when it was developed, they tried to include, for example, Indigenous knowledges and they spoke to Quilombistas, Quilombos are like way back in the colonial era, there were slaves that were able to fugitive slaves that kind of reorganized themselves in little Quilombolas, that was the name for it, slave villages, if you want to call them like that, and yeah, lived there freely and until today there exists this Quilombola culture. And in the development of the Brazil's AI strategy, they were all included, they were able to participate and their voices were heard. But in the end...
Figueira, Garrote and Zanatta at least, critiqued that Brazil's AI strategy ignores all Indigenous cosmology and replace the idea of colonial developmentalism, where the technological development is the primary goal.
It's kind of starting with a try of being different and creating a new way of thinking of the internet and AI in particular. And then at the end though, reproducing what others do, do I understand, right?
Yes. I think that the current global system, the dominant narrative, reproduces...
reproduces a certain way of looking at AI or of looking at technology and I mean this is not only true for AI but for other technologies as well and that's where I think my kind of I would say countermeasure or the ethnographic idea comes in to really look at the margins to look at the borders to look at the marginalized and to see what do they do and what do they believe and how do they frame AI or the AI of development? To open up a space for discussion, but also for reorientation for reimagining futures that otherwise are quite dominantly shaped by the narrative that is that we see right now nation states,
austerity, authoritarianisms. So quite bleak, I would say. I know that Fergal probably won't agree because you're the optimist. But yeah, quite a bleak situation, one could say. But it's interesting to see and to go to the like the roots, to the grassroots and to see what people do and how Indigenous people, for example, in Brazil, reimagine AI and society.
Chapter 5: How The Research Was Undertaken
Okay, so let's free the anthropologist that is in you. I would be grateful if you could actually share with us how you research this topic. So also in relation to what you have read in the context of philosophy of technology and theories of technology and the internet also on this, the colonial perspective, help us to decolonize
our imaginary.
Yeah, so maybe I'll start with the...
Yeah, it's not so easy to paint this picture of what I read, because as you guys, I come from the intercultural background, right, with our own literature and canon, and then I really try to find myself, or renew myself in the literature of the STS, for example. And what I found really helpful was the idea of assemblage thinking. I mean, that's surely nothing super new, but I found it very helpful. I'm talking, for example, about Barad, but also Latour, DeLanda, those are kind of the authors that inspired me. And in the end...
it's all about trying to see the world in relation, in relations that kind of span, could span through wide areas and include very different actors, people, tools, infrastructures, but also ideas that are all interwoven, intermingled. And it's not easy to separate them, and it's not about separating them, but to think them in these networks and how they co-produce each other. This co-producing idea then comes in from Barad, for example. And all that informed my thinking, I would say.
And then it was especially helpful that another guy, Lindgren, he introduced the idea of the assemblage AI or the AI assemblage.
And here, kind of, it all comes together, my whole thinking. And because when we look at AI as an assemblage, as, for example, also Kate Crawford does in her Atlas of AI, which I also can only recommend. And it appeared, funnily, just last month in a Portuguese translation in Brazil.
Yeah, so then AI is both embodied and material, and it is made from natural resources, fuel, human labor, infrastructures, logistics, histories and classifications, as Crawford would say. So then we look at AI and we try to study AI. We would have to look at the material, technological components, such as servers, computers, silicon chips, but also natural elements, rare earths, electrons, water.
And, for example, Leonelli, Sabina Leonelli, which I also try to read, she would say that institutional governance as a whole element in this AI assemblage is also important, but also social practices like coding, prompting, ethical discussions around it, and social technical imaginaries. And these social technical imaginaries that are a concept that stemmed from Jasanoff and Kim in 2015 already. So this is, I would say, the theoretical background that informed my then ethnographic work, very classic ethnographic work, where I did laboratory ethnography, organizational ethnography at the C4AI, the Center for Artificial Intelligence, in São Paulo, Brazil in 2022, yeah.
I hope I could answer your question, Luisa.
Sure, it's a start at least.
Maybe I can go into the ethnographic fieldwork that I did because as an anthropologist I'm very much interested in the practices I can observe and how people then really relate to each other and to the technology in the field.
How you work as an anthropologist in relation to AI.
Yes, so I was lucky that I could be part of the C4AI workforce and participate at work meetings. I drank the typical cup of tea with the researchers, more than one of course. And yeah, really try to participate in the space that is the C4AI. They have a very quite modernist building in the middle of the
University de São Paulo campus. It's in the Butantã district of São Paulo. A very nice building made out of glass and inside there is this playground one would say. So they had this idea of giving free space for the researchers at the C4AI. There is a couch, the typical elements that you would expect for example from a Google headquarter.
That was the space and then I tried to also do interviews of course, expert interviews, and I asked the researchers how they imagine the future, how they imagine AI, how their work is, their worries, but also quite mundane things one would say.
And out of that, I kind of created then ethnographic vignettes. I don't know, do you want to hear one?
Yeah.
So I was already back in Germany at that time. And there was a Zoom call I participated. And then
one of the participants said, yeah, we have this new supercomputer that arrived at the C4AI. And everybody was super happy and they were cheering and saying, yeah, we got it. And soon after that, they renamed the supercomputer and called it Arandú. And Arandú is the name for the god or the entity of wisdom in the belief system of the Tupi-Guarani. And I found this very interesting how they tried to incorporate this foreign machine, one could say
in their own epistemological world and make an accomplice in the research out of this foreign machine.
So it was, to be more exact, it was a DGX H100 from NVIDIA. So that really is one of the most powerful tensor core GPUs that are out there right now. And yeah, and it's very interesting how they tried to incorporate this new machine into their assemblage, renaming it. And in this renaming, there is a lot, right? There is a lot of cosmology shining through, a lot of idea about what it should serve, the preservation, for example, of Indigenous languages and knowledge.
Yeah, so that was that was another vignette from my field.
Chapter 6: Metaphors of “Devouring” and “Eating”?
Maybe the metaphors you use, I remember your presentation at our last conference, you were using metaphors of eating and devouring. I'm wondering if it's kind of relating to this, like, I don't know who is eating who,no what kind of imaginary is eating which other, but maybe you can say a bit on this, because I think it's super interesting in relation so eating and devouring in relation to writing also on AI. So why is this important? How has this been challenged? And how does it relate to an intercultural sensitive AI or indeed an Indigenous AI?
Tell us about it.
Thank you very much for this question, because now you're at the core of my own work or interests or what I found most interesting. So first I want to start by thinking about how AI or technology is narrated.
And when you look out there, you see a lot like the talk of systems are fed with data, there's data hunger. So the whole theme of eating is already out there. For example, Suleyman in The Coming Wave, he was one of the co-founders of DeepMind.
He wrote a book, The Coming Wave, and there's a whole chapter titled, “AI is eating the world”. And he says that, quote, “everywhere you look, software has eaten the world”. And Erin McElroy, they writes also very brilliantly about Silicon Valley imperialism as a global condition.
And they writes that it devours people's intimate lives, local epistemologies, and personal data. So we have this eating, devouring metaphor out there, and it does kind of it does a lot of work, right?
It's not just a description, because this metaphor shapes how AI is allowed to appear and what kind of force it is assumed to be and what kind of relation we imagine with it. And in this imagination, there is a passive world, right, that is devoured by AI. It's quite tech deterministic, one could say.
And what I encountered in the field is kind of the inversion of it, the active, interesting inversion as an anthropophagic AI. So the back eating or the eating back.
And this metaphor appeared to me because I am until today part of a WhatsApp group and there's also Paola Cantarini in there. And she has a very interesting project which is located at the Academia de Lettras da Bahia. It's called Tropic AI. And there she writes about anthropophagic governance and anthropophagic AI. And I would just quote quite quickly how she sees how she writes about it. And she says.
So, instead of, I quote,”Instead of singularity, multiplicity is affirmed by means of a necessary leap ‘from Amazon to the Amazon’— much more than a simple forest, since it is a question of imagining an anthropophagic artificial intelligence (rescuing Oswald de Andrade), or tropicalist, in the sense of developing an inclusive, democratic AI, multicultural, multidimensional, and focused on the Epistemologies of the South.” End of quote.
So, and there's a lot of in there, right? There's this world play from Amazon to the Amazon, the epistemologies of the South. Here she is relying to Bonaventura de Santos, but also Mignolo, for example.
And she says, she's trying to rescue Oswald de Andrade. Now here I would have to go a little bit back.
You're going to have to go back there, I think.
I have to go back, explain a little bit more about the anthropophagic movement. And it all started in the 1920s. There is a little legend to it. So there was a banquet in São Paulo. And Oswald de Andrade, he is author, very important figure in the modernismo in Brazil. And they were having a banquet. And his wife, Tarsila do Amaral, she's a painter and artist, was also participating. And they had frog legs to eat in this banquet. And Tarsila do Amaral said quite jokingly, Oh, this frog legs look kind of like man legs. And we are all quasi-anthropophagai.
And relying on that, Oswald de Andrade developed his first Manifesto Antropófago. So this anthropophagic manifest, it appeared in 1928 and a little bit before that, already Tarsila do Amaral painted a picture called Abaporu. And Abaporu is a Tupi word for the man who eats man. So you see the whole Manifesto Antropófago that Oswald de Andrade wrote was very important for the modernismo, for the whole modernism movement in Brazil.
It is written in a very ironic tone. There are a lot of puns in there, for example, from Hamlet. He reinvents the Hamlet phrase to be or not to be into Tupi or not Tupi. And Tupi is this Indigenous language group.
Ah, okay, sorry. I understood something else there.
Yes, I know. It's easy to misunderstand. Exactly. But Tupi. Exactly. Or, for example, another passage.
Ah, yeah, exactly. For example: The naked man is necessary. Or the nude man. It depends on the translation. But we don't know when this nude man will arrive. The American cinema will introduce him to us. So it's a very contextually rich piece, the Manifesto Antropóago. And now we have to rely a little bit on the history, what happened with it. For example, Viveiros de Castro said that it was actually a decolonial philosophy avant la lettre.
And it could, for example, Garcia writes that it is a diagnosis for society marked by colonial dependency and a therapeutic remedy. And the whole idea is that foreign elements and influences and culture should be consumed, eaten up, metabolized, and turned into something completely new in the search for Brazilianness. As they put it, for example, Dunn writes about it, or Sequeira. There's a very nice translation from the 1990s by Bary, and she writes that in the end it's all about creating a muddy copy. So the idea behind that is that from Europe at that time,
what arrived in the former colony, Brazil were pictures, music, that all seemed very clean and modern and new. But it encountered a reality that was far from being easy, clean and modern. There was the Amazon, there were the Indigenous, there were the former slaves. And all of that inspired Oswald de Andrade to write this Manifesto Antropóago and the idea of devouring those elements and digesting them into something new.
Chapter 7: Ideas and AI as Lived Reality
Okay, that's very, very interesting. And okay, I understand that on a narrative level. It's about an imaginary, a way of thinking about AI and a way of creating meaning and an attempt to decolonialize. But this is ideational, isn't it? These are ideas.
So how does this relate to AI as a lived material reality in Brazil? Is it still intertwined in a type of digital colonialism, as in to the US, possibly China dominate here, actually? And this is just sort of an attempt to rethink it or how does this relate? I suppose this is my question. This idea, very, very interesting ideas in terms of an imaginary and with a rich sort of intellectual history, as I understand it. But is this just attempt to reinterpret and rethink? And how does it relate to the material reality? I suppose this is my question.
That's a very good question. So in the end, how I try to use this lens of the anthropophagic AI to make sense of the data I gathered.
The idea floats in the air there at the C4AI, but what I encountered materially was of course there was a huge dependence on, for example, the Nvidia computer, on codes that were developed in the US or in China.
The money, the whole structure of the C4AI relies heavily on funding by IBM. And IBM is very active in Brazil since the early 1960s, I would say, even before that, and trying to stimulate development there. So the C4AI is very dependent on influx from outside, money from IBM, the technology from Nvidia from the US.
But what they do there then is, at least that's what I observed, is they try to apply it to the problems that are very much rooted in the Brazilian reality and the Brazilian soil. For example, there's one project line that's called AgriBio. So one could say it's all about Agribusiness.
And here again you have, that's what intrigues me, you have again this idea of eating and supplementing stuff. And then, for example, this camel project tries to tackle the hunger problem in big cities, trying to gather data, analyzing where could be the next outbreak, for example, of hunger, and developing smart systems, AI systems, that predict, for example, this hunger outbreak. So, yes.
So, of course, you're right, it takes place more in the realm of figures, of imaginaries. But if you look at it materially, I would also say that this idea of digesting, this idea of taking the foreign, for example, computers and codes and help and money and then kind of metabolizing it into something that could really help the people in Brazil. Still it reminded me of this anthropophagic idea of the anthropophagic governance.
But I have to give you, right, it's a bit, it's a stretch, it's a lens I use to make sense of the data. And it is aspirational, right? It's an ethical approach.
But I mean, at the same time, if the programmers are internalizing these ideas, then this will also be expressed in how they program and what they do in terms of the development of Brazilian AI, even if everything is financed by IBM or whatever else.
Chapter 8: Anthropophagic AI Anthropophagic AI
Exactly. And one another important part is that the antropófago, the anthropophagic figure or man or people, they are thought of as an anti-myth, an anti-mythical device. That's at least how Nunes in the 1990s put it. And you see, when it comes to AI, I think there's a lot of mythological speech. They're opaque systems and they're future-oriented all the time, I would say, when we speak of AI, we think of automated
situations, systems, machines in the future. And what the anthropophagic AI would do is to ground it again, to muddy it, to break it down, to bring it down to earth. And that's what makes it an idea from the south, right? And that's what I encountered there in the C4AI with the researchers. They are not developing grand utopias. They're not thinking of shiny futures. Sometimes they were even saying, it's impossible to imagine something, something new, something good, because we have to take care of the really urgent problems we have here. And this is, I would say, very particular. That's a perspective from the south. And therefore, one could say a decolonial.
So the third way is a pragmatic AI.
Yes, it is.
It is kind of a pragmatic AI. Yeah, one could put it like that.
But also it's kind of
the creation, going back to our interest in the concept of interculturality, is the creation of an interculturality, taking what is there and trying to do something new with it.
Yes, exactly. There's a lot of creativity in there, right? And sometimes, I think, at least in the past, as interculturalists, we tried to focus a little bit on the aspects of harmony, of seeing intercultur al encounters, of course problematic, but at the same time fruitful. What I think what the anthropophagic idea does is to show that it's more about friction, and sometimes it can be even hurtful or violent, right, in this metaphor there's something violent, of eating, consuming. But also, at the same time, a transformation, right, a metamorphosis, be it epistemological or material.
And I think that's the intriguing part. That's what fascinates me. Yeah, so in the end I tried to come up with a definition of anthropophagic AI, kind of relying a little bit on Cantarini's idea but then also developing it. Do you want to hear it?
Sure!
Okay, so in the end I would say that anthropophagic AI are AI systems that adopt so anthropophagize, incorporate the semantics and discursive material realities of the tropics and making them the basis of their own machine learning and decision making in order to deal with specific problems and issues of the global south. That I would say is the core idea of the anthropophagic AI. Maybe it's too technical again but that's what I would...
That's what I would try to, if I had to put it in a definition, that would be my definition.
Yeah. And I don't think it's too technical at all. I think it's very clear and very full of multi-perspectivity and multi-dimensional, let's say. Because it's not just about technique, it's about what you do with it. And that has to do very much with culture. So thank you so much. Muito obrigada.
De Nada. [You're welcome.]
It was very, very, very enriching. And how do you say? Enriching as a meal. A very challenging meal, I think. I will need a bit to digest it, but it was a very inspiring meal. So I will remember this
for a long time. It really was very inspiring for me. Thank you so much, even if I was listening to you already. Every time we have a chat and every time we think of things, we can see new dimensions. And now you presented also different aspects. So thank you, thank you, thank you very much for that.
Thanks from me as well, Emilian, and we can go and devour some data now, I think.
That's a good idea. That's a very good idea. Yeah, thank you guys so much for having me. It was a great pleasure.
Thank you, Emilian.
Take care.
So thank you very much to everybody for listening to us, staying with us, and check out the other episodes we are producing. Then you can...
read more about this topic and even listen to a presentation on our YouTube channel, in our website redico.eu and also you can connect with Emilian Franco, with us and with many other researchers, but also experts from civil organizations and so on on our hub, hub..redico .eu.
So we are looking forward to getting in touch with you.