Anthropology and Artificial Intelligence

Since Child Data Citizen was published in 2020, what concrete actions have you seen in how the social media companies collect data or how governments protect our rights?

I’ve seen different types of transformations so much so that over the last year I was contracted to write a new book, The Children of the Algorithm. I realised I needed to write something new. I still had a lot of data from the Child Data Citizen project that didn’t fit in the first book, and a lot of reflections and theoretical consideration of how children were being datafied.

But also the pandemic, from what I’ve seen, accelerated massively processes that were already in place; to give an example, I had to, after five years of researching children and datafied citizens, and I had no choice, to create a Google Classroom account for my daughter.

So there was this situation into which families were thrown, and most of the time we didn’t have a choice. It happened really fast, especially because governments and schools institutions in general implemented those transformations more rapidly and more extensively than they had done before.

And also what we saw in very short period was the expansion in the ways AI technologies were being used to profile people. Again, going back to the education sector, but also thinking about the workplace, we’ve seen the use of different types of software that analyse facial features. For example, Proctorio which was used to make sure students didn’t cheat during exams.

Why would an anthropological approach be different from say, Virginia Eubanks, who uses ethnographic methodologies and has a real context-specific understanding of what’s happening on the ground.

And, of course, there were huge social, political, and critical implications in the use of these technologies.

On the positive side is that the debate about algorithmic bias, the adequacy of these technologies, and also the importance of protecting children’s rights has expanded. We also saw some key steps such as the white paper by the European Union for the regulation of artificial intelligence. It was kind of a new world that was happening over the last couple of years. I decided to start again and write about it.

What do you believe anthropology uniquely brings to the study of data technologies and AI, and why do we need more anthropologists working in this space?

This is actually very fascinating question. When you sent me your questions, Dawn, I was reading over them, and this was the question, right? How am I gonna answer this? So why would an anthropological approach be different from say, Virginia Eubanks, who uses ethnographic methodologies and has a real context-specific understanding of what’s happening on the ground.

And my answer to that is, first of all, I was lucky to be supervised by David Graeber, in the last year of my PhD, and I was his teaching assistant for two years. And I think most of the things that I identify with anthropology, I identify with David. I’ve always thought if I have to think about what the values are, I think those are the values that I would be looking for in an anthropologist.

But I’m also thinking about other people who inspired me in my career such as Tim Ingold (whom I’ve never met). There’s a particular article from 2014 where Ingold questions what is ethnographic. He gets really frustrated about the fact that when we, as anthropologists, review articles or funding applications it’s stated this project is going to entail ethnographic interviews. And you’re thinking, well, I don’t know what makes an interview ethnographic, and what doesn’t it make it ethnographic.

So, in terms of these kind of debates about what anthropology does, I am very much inspired by these two figures and their ways of thinking. And to be honest my belief is that there is a fundamental difference. After the years and years I’ve been working across the disciplines (because I built my career in media studies or communication studies and my PhD is in social anthropology) and struggling with these questions, I think that what makes anthropology unique is that, in other disciplines — like sociology, political science, media studies — ethnography is often, too often, used as a method. And as a method that usually it is about exploring the qualitative dimension of life.

But anthropology, ethnography is not only method, it is an analytical perspective. You ask your question through your ethnographic engagement.

It’s how people negotiate that which has a lot of similarities to the ways in which it is used in anthropology. But anthropology, ethnography is not only method, it is an analytical perspective. You ask your question through your ethnographic engagement.

The level of ethnographic engagement that you might have had when you were a PhD student is very difficult to get once you advance in your career because we don’t have the time to go away for a year. I have two daughters, I direct an institute. There’s no way that I can actually do a year of ethnographic fieldwork.

And so what anthropologists usually do is they use their analytical, ethnographic lens in everyday life. And I think that that’s what makes us unique, apart from the anthropological theory. Often when we use this analytical lens, we also engage with anthropological theory that has been fascinated and culturally-specific, and it’s rich in the engagement with the problems that surface from everyday life.

For instance, the anthropology of personhood was one of the theories that really inspired me when I was thinking about how we were datafing citizens, or the anthropology of value. That’s what makes anthropology unique. And that’s the very positive side of anthropology.

What do you think are the most pressing areas for further research by anthropologists in this space. Where should we be focusing our gaze in your opinion?

With topics like artificial intelligence and data and algorithms, you’re seeing — even more so I would say than when internet technologies developed — much, much more anthropological engagement. The Royal Anthropology Institute, for instance, are doing an entire conference on artificial intelligence. And I thought, oh my.

I was doing my PhD in social anthropology between 2006 and 2009 and at the time I was very much engaged with the key questions about media anthropology or digital anthropology. It was a very small niche of anthropologists working in that field, the EASA media anthropology network, which was launched by John Postill in 2005. (I chaired the seminars for many, many years, but I resigned last year because I had done it for too long.) There were a lot of media anthropologists around the world but media anthropology was still something that anthropologists were not really engaging with was with AI and data.

AI and data technologies are involved in deeply problematic and hierarchical processes of meaning construction and of power-making. There’s no way that anthropologists could shy away from this.

I think we’ve actually seen a reality check for anthropologists because these technologies that filter so many dimensions of social life. They have also become agents in many ways and involved in deeply problematic and hierarchical processes of meaning construction and of power-making that there’s no way that anthropologists could shy away from this. I mean, there was no way that they could. I’m thinking about the people who have always worked on these topics, but there’s so many people, there’s so many anthropologists who are picking up from an ordinary understanding or really expanding it theoretically. That’s very interesting.

Boellstorff, for instance, in 2013, published an article, which I’m super fond of, on big data where he traces back, and this is very interesting, this idea of the algorithm, and how anthropology relates to algorithm, to Lévi-Strauss and where he talked about ethnography and ethnographic algorithm because through ethnography you’re effectively…ethnographic method, that’s what we do, right?

I wrote an article on how Graeber’s theory of bureaucracy, violence, and value enables us to understand surveillance capitalism.

So there’s so much really inventive anthropological theory that we can use to learn about this current time we’re living in. Obviously, I’m biased but one of the best books, that really, I think, makes us see things for what they are and, is David Graeber, the Joys of Bureaucracy [The Utopia of Rules: On Technology, Stupidity, and the Secret Joys of Bureaucracy], published in 2016.

In 2021, just after he died, I wrote an article on surveillance capitalism, which is an homage to David. It’s published in the Annals of the Fondazione Einaudi, where other prominent anthropologists have published. And strangely enough, it was a special issue in honour of Sahlins. I was asked to write this article about David. I wrote how his theory of bureaucracy, violence, and value enables us to understand surveillance capitalism.

And he talks about how bureaucracy is actually there to construct social truth, but effectively, this type of bureaucratic work has been now replaced by algorithms and artificial intelligence. I tried to make that connection in my article [David Graber, Bureaucratic Violence and the Critique of Surveillance Capitalism], to try to really show why it is important to look at anthropology and why on the other hand anthropology engages with these reflections about what’s happening, and what has happened over the last couple of years.

Your book and TED talk are fantastic examples of how anthropologists can make their work both academically rigorous and accessible to the public. Does anthropology need to get over the idea that writing for the public is something to be ashamed of and also that such writing lacks scholarly rigor?

It’s a very difficult question in the sense that I can only talk about my personal experience. I did the TED talk and I did try to push the Child Data Citizen findings out there a bit more than I used to do in my research (such as things I wrote on time and temporality, very theoretical, that nobody really wanted to read).

But I did it because I thought it was my responsibility, that I had to do something. I was seeing things across the two countries, the UK and the US, that I was dividing my time between. I wrote Child Data Citizen across two cities, Los Angeles and London. What I was seeing was that, especially in countries like the UK and the US, the datafication of citizens was exposing society to all sorts of inequality, of systemic bias, of injustice. (I was most certainly not alone, there were many scholars who were doing really interesting work.) But it was important to get the message out there in terms of campaigning and raising awareness.

And that’s how it happened really. That’s how I got to do the TED talk. I also realized very quickly, both in the US and the UK, but in Europe as well, and now in Italy, that I was lucky enough to have a research question that people had not thought about it. The experts in children rights were already working on some of this, of course, but it was the broader society that wasn’t very much thought about.

I felt I needed to be more active in society. I owe that a lot to where I come from academically.

And when we were talking about children and data, we were mostly talking about sharing things, the so-called parents that share information on social media. I thought that it was so individualistic an approach to blaming parents for a very, very, very tiny practice out of the world of data collection so that I thought that I needed to try to become more public.

I was really lucky to meet anthropologists like David Graeber, but I’m also thinking of Brian Morris or others I met who were active in some way, or were trying to be. It was all about engaging with the topic publicly. I felt I needed to be more active in society. I owe that a lot to where I come from academically.

I don’t think Graeber did it deliberately in a sense that he saw himself as being an activist, but the way in which Graeber influenced public discourse, the reach that his book had, it was very public. People from across the political spectrum were reading it. And I think that’s great. That’s what anthropology should do.

But it is true that, within the discipline, we tend to believe that if you actually are public, then the type of knowledge that you’re producing is thin or that you’re objectifying your findings. Which to a certain degree is true. I mean, to be honest, summarizing what three or four years of ethnographic work in a 10-minute TED talk? We can’t lie to ourselves, it’s true and it’s happening. But we can live with both. It’s not that, if I do a ten minutes TED talk then I can’t dwell, like I do in this last book that I wrote, on profiling and categories and Mary Douglas. We can still do that. And I think the world can only improve from that type of anthropological lens. It’s very important.

A critique of anthropology

One of my major critiques of anthropology (there is actually a journal called Critique of Anthropology) for young scholars is that anthropologists often shy away from engaging theoretically with disciplines that do not share their approach.

There’s a good reason behind that, which is that I think anthropology as a discipline is particularly ferocious with its disciplinary boundaries because we are smallish discipline, because we do have a unique perspective, because there are not many anthropology departments or academic positions in the world. Many anthropologists simply go into all sorts of departments such as Human Geography and others.

Anthropologists often shy away from engaging theoretically with disciplines that do not share their approach.

So there’s a commitment for those who have to flee from the traditional anthropological department to maintain that sense of identity, that sense that we are doing high-quality research, that has a different dimension to it, that has something different, and we all know that it has something different.

There’s a positive element or a justifiable element to this critique. But over the years, I have often personally encountered a great deal of anthropological snobbism, a great deal of, oh no, we don’t talk about this.

Mind you, for someone who was co-supervised by someone in media communication and someone in anthropology, I used to struggle because every time that I brought back a chapter, which was completely decent chapter, I would get, “this is not anthropological enough”. I remember that from my PhD days: this is not anthropological enough. It’s not true.

At the moment I’m a visiting professor in Münster, and I teach a course on visual anthropology. It’s quite interesting because we talk about ‘what is anthropological enough’, what makes something ‘anthropological enough’. I would say to people, to students, read. Read plenty, read as much as you can. So if you’re interested in a particular topic, especially like artificial intelligence, you’re going come across so many books that are really high quality, but do not share an anthropological sensibility.

Treat it as an anthropological experiment. Read these books to understand what our cultural values are, why as a society we are talking so much about artificial intelligence.

But it’s important to engage with those books. It’s important to see what’s out there and not be shy and not be a snob if you want to know things. Don’t be critical or pre-emptive from the beginning. Treat it as an anthropological experiment in the sense that you read these books to understand what our cultural values really are at the moment, why as a society we are talking so much about artificial intelligence.

The other aspect, when we think about artificial intelligence, is that we’ve never been in this situation, not in this way, where we have to teach these machines to read our world. We have to teach them basic concepts, like what is a child, what is a woman, and so on. And obviously, as anthropologists, we know that these concepts are so, so complex, they are so culturally specific, they are so biased.

This is something anthropology can do: highlight the way in which these technologies are always inevitably going to get and be biased. There’s nothing that we can do about it. There’s no way to fix them. There’s no way. But what we can do is to limit that bias, or engage in proper conversations about what type of definitions and categories we’re teaching these machines.

We’re going to see such great research emerging in the next few years. I’m actually looking forward to that.