Essec Business School

With Cristina Alaimo and Lauren Waardenburg

 

Julia Smith, Editor-in-Chief of ESSEC Knowledge: Hello everyone and welcome to be in the Know, the ESSEC Knowledge podcast sharing the research and expertise of ESSEC professors. Today, I'm here with Cristina Alaimo and Lauren Waardenburg, both professors of Information Systems, Data Analytics and Operations. They're also the founders of Theorizing Data and AI, a community dedicated to scholars exploring the impact of AI on organizations, society, and institutions. Fresh off the fourth edition of the community's annual conference, they're here today to share their insights. To kick off our conversation, can each of you tell me a little bit about your research?

Cristina Alaimo: Hello, Julia. Thank you so much for inviting us. I’m very happy to be here talking about Theorizing Data and AI. I have always done research around data from a social science perspective: asking what data are, how they came to be data, what they represent, and the impact they have on how organizations make decisions and on how organizations come to represent the world they know.

I think that's a very important connection with AI because in the majority of research data are treated as input to AI systems. AI is trained by data, but data also have a history on their own. They are also the output of very long and complicated technological systems, organizational decision-making, and even historical path-dependencies. So that's my link between data and AI and that's how I came to study AI and be involved with Lauren on this project.

Lauren Waardenburg: That's a nice transition. Hello Julia, and hello everyone who's listening. My name is Lauren. I have been studying AI for quite a while now, since 2016. I cannot believe that it's already been 10 years already! I study AI mostly in extreme contexts meaning the police, military etc. I'm an ethnographer. That means that I actually go into these organizations to study how this technology is changing their work. That's how I also very quickly learned that studying AI is impossible without studying data. So these technologies, you cannot see them as separate technologies without understanding what a massive consequence data has on everyday work, not only in these extreme contexts but beyond that. I study that from a sociotechnical perspective, looking at both the social consequences on work and understanding the technical details of these technologies and how they change through our social actions.

Julia Smith: Thank you so much, Cristina and Lauren. It's clear that you both have a keen interest and a lot of expertise in AI and data. Can you share with us now how the Theorizing Data and AI community came to be?

Lauren Waardenburg: Absolutely. That was in the summer of 2022 when Cristina and I talked about this for the first time at EGOS, an annual conference. Cristina and I had already known each other for years. We always ran into each other, because we are interested in the same topics, but the interesting thing about it was that we were in two communities that did not really talk to each other. We were interested in the same thing, but not having an ongoing conversation about data and AI. We said to each other, “How nice it would be if we brought together our two communities into one where we can actually share these ideas!” 

That was the initial thought, which grew very quickly into, “Well, if we bring our two communities together, maybe we should have our communities be inspired by even broader communities and different fields.”

That's how our initial idea, which was then still a workshop, came to be. We applied for some funding back then from the Society for the Advancement of Management Studies. We got that for the first year, which helped us to set up the first version of the workshop. The great thing about Theorizing Data and AI is that we brought together people from completely different communities. We are in information systems, and management and organization studies are close disciplines to our own. A bit further away are social sciences, accounting, and computer science. We try to bring in those communities that are in our orbit. We invite people to come and give a keynote so that we can inspire and be inspired by each other. The first round was very small. I think we were with about 50 people back then and from then it's been growing and growing!

Julia Smith: It's great to see you bring so many different researchers with so much varied expertise together. That makes me curious: how do you bridge the gap between social science and AI and data in your own work?

Cristina Alaimo: That's a challenge of course and something to which we are very committed to. We do it in a couple of ways. We do it in our own work. This is also one of the scientific spines of the community and the conference. In my own work, this has been sustained by a relentless attention about not taking data for granted. I inquire around what data are: 

• How are they produced or generated?
• What are the implications of it?
• What are the assumptions that go into data?
• What kind of truth does data reveal about the world? 

If you think about how our organizations now are data-driven, you really understand the importance of integrating what then with my co-author Jannis Kallinikos we call “social science of data” into the study of data technologies, among which of course, is AI which is today basically sustained mostly by data. 

This was something we wanted to do since the beginning. We wanted to facilitate discussion around data and AI and not to take these technologies for granted, but rather to reconstruct the histories, the decision-making, the assumptions, the technologies behind data and AI. That's what we think of as a social science perspective to data and AI, which doesn't mean not to look at the technologies or the impact of these technologies, but rather opening the black box of technologies and at the same time also inquiring about the history and the organizational context where these technologies are shaped.

Lauren Waardenburg: I think that one of the key elements here is also to step beyond the “hype”. I think our initial version of the workshop was also called “Beyond the hype of AI”. When we did the first version, it was still what we now call traditional machine learning; now we have generative AI, so now the world is different. 

It's very easy to fall into this hype of both what technology can do and how the world is going to change, etc. If you follow that kind of reasoning, you overlook a lot of nuances of what the technologies actually can do, what these technologies are actually doing, and what we are doing in our relationship to technologies like this. That was generally our aim for Theorizing Data and AI to push against that and really start to unpack and theorize what it is actually that we talk about when we talk about AI or when we talk about data. To translate that, because the question was how do I do that in my own work, I hope that I also do that in my own work: not fall for this “AI is going to change the world and we are just passengers in this and let's see” but instead, really dig into what is changing, what is still staying the same, and how are we influencing it, what are the kind of activities that we are constantly putting into this, for example.

Cristina Alaimo: If I may add one consideration to what we were discussing, I think it's also very important. Perhaps we are saying something that sounds theoretical, and just like academic reasoning, which it is of course. We are talking about an academic community. But there are some ways in which this can be operationalized into concrete discussion and can be made concrete: facilitation of dialogue across disciplines which is one of the tenets of our  meetings and across people. For instance, let me give you just two examples.  One is to be very focused on the phenomenon. What do we mean when we want to study data and AI? Lauren mentioned it: there is no AI without data. Now machine learning, the traditional predictive model and increasingly generative AI, or this probabilistic model, do much of the work with data but also there is no data without AI increasingly. So then when you start talking about this stuff and you start calling other people with different backgrounds to reasoning around this phenomenon then you can concretely expand the discussion around data and AI and avoid taking both for granted or fall into this hype narrative.

Julia Smith: Thank you. It sounds like there must be some really interesting conversations that take place at these conferences. What were some standout moments or conversations from the conference this year?

Lauren Waardenburg: That's the thing we are most proud of, I would say, is seeing this community evolve and also the conversations that happen and that keep on going basically also after the conference has finished. So there are two things that I would like to highlight based on what came out of the conference. 

1. What we managed to expand over the years is what we now call the pre-conference workshop. That is a workshop dedicated to early career scholars most of the time but also very early studies in which people can present their very early ideas on AI and discuss a whole day only in round tables, discuss these ideas together to develop. The only goal that we have that day is that people come out and say I really benefited from these discussions today, I really felt like something opened in my head about this. and this grew from like the first version of the workshop was almost only like this but then it grew and you cannot do that anymore. So we needed to organize something else. So now we made this into a whole new added day at the beginning. It's something that energizes us so much, to see how people work together and discuss and are involved in each other's topics, This year we also had a fantastic keynote by one of our colleagues, Lior Zalmanson who talked about the role of art: the role that art can play in our theorizing about AI. 

2. The keynotes are the second thing that I wanted to mention because that's the thing that gives a lot of energy every year. We can bring in people with different insights and different perspectives on this topic and by doing that, the whole aim is to trigger something: we don't want to hear what we already know. What we want to hear is something else, and what we see now year after year first of all is that we make the keynote speaker sometimes very uncomfortable! Because we ask them to come in and speak to a community that is not necessarily their own and then they're in front of this community and they feel very vulnerable sharing their ideas but almost all of them have have come out and said “This was so nice that I could actually do this and that we could interact”. This vulnerability, and the willingness of these keynote speakers to be vulnerable, leads to a funny thing: these keynotes travel, so they trigger other initiative, other ideas, and other conversations that keep on going either offline or online. That's super cool, and that's definitely a highlight not only this year, but every year.

Cristina Alaimo: So this year in London, at the London School of Economics and Political Science, our latest conference had 140 participants, three keynotes plus one by Lior that was called a non keynote because it was about art, a panel… Then we have a scientific committee which is formed by all the different keynote speakers that have participated in the conference throughout all its editions. So we have 20-plus professors: of political science and sociology; public policies expert, Helen Margetts for instance who is now the upcoming director of the LSE Data Science Research Institute; Elena Esposito, a sociologist; Mike Power, a professor of accounting; Jannis Kallinikos, Youngjin Yoo, we have people from all over Europe and our disciplines, but also management. We have ESSEC people, we are not the only two involved from ESSEC! There’s our colleague Harris Kiriakou who is also a part of the scientific committee, and we also have professors Lars Andraschko, Christoph Mueller-Bloch and Alentina Vardanyan, and we hope to have more colleagues after this podcast. So hey, you are all invited! We really hope that we can spark a conversation around these themes. 

Something that is particularly relevant is that these conversations have also crystallized in scientific projects and this is something of which Lauren and I are particularly proud. I would like to mention two projects. 

• We have an upcoming editorial on the European Journal of Information Systems. Lauren and I are co-authors with Jannis Kallinikos and Youngjin Yoo. This will be really the spine of how we can theorize data and AI. I would like to stress the fact that the editorial will make the effort of making a difference between machine learning or traditional predictive AI and deep learning. What does that mean for how our organizations learn? Do we need to treat these two technologies, what we now all call AI, without making too much of a difference in different terms, when we study their organizational implications for learning and for knowledge-making? That's one of the questions we will ask ourselves in the editorial. We hope to provide answers or at least generate more questions to our colleagues. 

• The second initiative is a special issue. We have published the call right now. The special issue is open and is on theorizing the data and AI nexus. It's a special issue of the European Journal of Information Systems and it aims to be interdisciplinary. The most important thing for those that are interested is please send stuff on data and AI: not only data, not only AI, try to understand how your own research, your own background, your own discipline, try to theorize about the nexus between data and AI.

Julia Smith: The news of your special issue and your editorial has made me think of another question. It's a bit of a tough one, but based on what you've seen from your own work and that of the community, what are your predictions for how we will interact with AI in the next few years?

Cristina Alaimo: We are in the business of studying the past and the present, so it is always a bit more difficult to study the future because it's not yet here! We can reflect on a couple of important points. 

• One I have already mentioned is that we need to be more specific about AI. AI is a family of technologies, it means so much, it includes so many different things so the effort we are doing for instance is trying to make a difference between machine learning and deep learning is just one concrete example of how most probably research will change when thinking about AI. 

• The second important aspect is that we as researchers need to make the effort of going beyond the individual level and start thinking more about the collective. By which I mean, we have studied a lot and heard a lot about the impact of generative AI on individuals. But what about the collective? How do we go up one level and try to understand how generative AI impacts learning, not just the learning of individuals but also institutions and organizations? So this is something that will likely change and will depend of course on our commitment and our engagement but it is very much needed. 

Connected to this, I want to close with a concrete idea or perhaps a call to action, which is to construct a dialogue with policy makers and regulators. Not just with practitioners, that of course we know, they need us and we need them, but we live in an era that needs rules and regulation that do not have to stifle innovation or to close to innovation but rather help this innovation travel. On this, I think we have a lot to say. Again, I call back to the social science approach to data and AI; this is not something that is only an economist's or lawmaker's topic. Not any longer. This is a topic for management researchers, for information system researchers, for people who know the impact of technology on organizations and society and can enter and participate in this interdisciplinary conversation. What happens if I forbid a 16 year old to use social media? These are not just economic issues, they are sociotechnical issues that need to be explored through and through.

Lauren Waardenburg: If I can add to that, I think especially for the policy makers there are a lot of taken-for-granted activities and roles in our work and in our society that for the longest time didn't have to be included or considered especially because for a while AI was a quite narrow application area. Now, we are seeing that so-called narrow application area has landed in our households and we're looking up recipes on it, etc. For example, we assume that a manager is there managing and because of that, we haven't paid a lot of attention to the role of a manager when it comes to the workforce and GenAI, etc. Now that's not a question anymore. We cannot overlook that anymore. We cannot assume that a manager just has a managerial role and carries it out the way that it has always been. So these are “taken-for-granted” or hidden consequences, new types of work, new types of roles that need to be included in governing and policymaking around AI and are often not included. It's very often only either about the worker or the technology and not everything that revolves around it and that keeps it going.

Julia Smith: Yeah, it definitely is clear that there needs to be a lot of people involved in the conversation around AI. To close off this conversation, I'm wondering if you have any advice for your fellow researchers or even those of us just exploring our own relationships with AI.

Lauren Waardenburg: I reiterate what you just said. There's a lot of people involved and since there are a lot of people involved, that also means that there are a lot of fields that we can learn from. That's also one of our core ideas for Theorizing Data and AI. Let's not stick to our own little cubicles and our own little boxes and then reason from that. Let's not say “theorize” and just try to understand what's going on with us from that little box that we're in, but instead bring in insights from different fields. Think about social sciences, think about art, think about accounting if you're a management scholar, etc. because these fields can show us sides that we haven't seen yet and that we might have overlooked. Get out of that cubicle, “transcend” like we like to say at ESSEC, and learn from different fields and different areas because they can surprise you with how much you can learn from them.

Cristina Alaimo: For our colleagues, I second everything Lauren has just said which is super important. Here’s also some practical advice, which is what we did together: find your own community. Do not just stick to communities that exist, but form it if it doesn't exist yet. That's what we did! We did it out of a need that we felt: Let's talk, we study two different things from two different angles. Let's talk, which is exactly what Lauren was saying! Let's bring together different points of view to better understand a phenomenon. This kind of phenomenon-driven theorizing which very concretely for us meant founding a community and a conference and then you know launching a special issue and editorials and so on and so forth, has generated so much for many people and particularly for young researchers, PhDs, postdocs. We are very very proud of it and we will continue working as we did so far. This is also the advice I think that I really would like and I think Lauren too, to give to colleagues. We need to talk, we need to share, we need to facilitate conversations. If you see that this doesn't happen through canonical channels like very big conferences or meetings, make an effort, create your own community or join a community that maybe is not your own but can be a community in which you find your academic or scientific home.

Julia Smith: Thank you so much, Cristina and Lauren, for taking the time to speak with me. With AI transforming society and organizations, it's clear that we need to take this type of interdisciplinary approach and it's really inspiring to see professors right here at ESSEC who are so engaged and committed to this process. Thank you.