Subscribe
Share
Share
Embed
Born & Kepler is named after the mathematician and scientists Max Born and Johannes Kepler. This bilingual podcast, offered in both German and English, dives into the expansive world of Artificial Intelligence (AI), exploring its foundations, evolving technology trends, academic search, and its impact on businesses and society.
Born & Kepler will feature a diverse lineup of experts from academia, venture capital, private equity, journalism, entrepreneurship, CTOs, and policymakers. Each guest offers unique insights into how AI is reshaping their sectors and what we might expect in the future.
Our goal is to provide a deep understanding of the core principles and breakthroughs in AI, enabling you to stay updated with the latest advancements in AI technologies and how they are transforming industries. During our episodes, we will explore how AI is influencing business strategies, optimizing operations, and driving innovation. We will also explore the ethical, social, and regulatory aspects of AI in everyday life.
Andreas Deptolla (00:00.979)
Stella, welcome to the show. It's nice to have you. Are you in Cambridge right now or back in Greece? Where in the world are we finding you?
Stella (00:11.5)
Andreas, thank you very much for inviting me to your podcast. I'm quite excited to have this conversation with you. I'm in Cambridge right now still for actually for another day going to Greece to celebrate Christmas tomorrow.
Andreas Deptolla (00:26.525)
Very nice, very nice. Thanks for taking the time now. I know it's a busy time of the year. Tell us about your journey getting to Cambridge, right? From an academic perspective, what motivated you? What were some of the challenging moments to get to where you are today?
Stella (00:49.454)
So when I was a teenager, it was such an interesting time because it was the internet era, right? And I remember connecting with people on the MIRC and going, know, visiting websites to get information, et cetera.
So I come from the tech background because I was such a tech enthusiast when I was young. So my first degree was electrical and computer engineering. was, I think around the time that I was ending high school, they had started incorporating some classes on developing algorithms. It was just in pseudocode, but I loved the challenge of.
Andreas Deptolla (01:22.653)
Mm-hmm.
Stella (01:38.068)
having to solve a problem with an algorithm and that drove me to studying electrical and computer engineering. But while I was at university, I also really liked interacting with people and that drove me to do a master's in business informatics. I wanted to go abroad and I found these wonderful masters in business informatics in Newtruc University in the Netherlands. So I ended up
Andreas Deptolla (02:06.472)
Mm-hmm.
Stella (02:07.79)
going to the Netherlands for a couple of wonderful years. It was an incredible experience because a lot of what we did there was to understand ICT entrepreneurship and get exposed to also how tech startups work. But a lot of it was also quite technical as well. And I remember...
I think my favorite class was data mining and my whole master's thesis was around data mining. ended up doing an internship in a software as a service organization with headquarters in Delft in the Netherlands. And I had to develop a prototype which was going to analyze the...
various usage data, cluster the users, as well as provide some predictive maintenance services based on all the log data basically that were collected. And what was interesting there was that I remember finishing my prototype, feeling very proud, I got an excellent mark for my master's thesis, you everything was amazing. then...
I was asking those wonderful people at the company, OK, so how have you started using this product? And they were like, no, we haven't had the time yet. And in a way, I was wondering. I thought that this was working well. What's happening? And this question of me being this tech enthusiast and thinking we can resolve everything with technology, how come people don't resolve their issues with technology?
Andreas Deptolla (03:27.485)
Mm-hmm.
Stella (03:49.554)
And that was my driving question in my PhD when I started. I wanted to understand, if we assume that the technology works technically, why doesn't it work in practice when you introduce it in the workplace? I was very naive. I was definitely a tech enthusiast. But that drove me to a business school. I did my PhD at a wonderful, wonderful group. It was my intellectual home.
the Keen Center for Digital Innovation at the Freie Universiteit Amsterdam with my supervisor, Marlene Heusman. having that data mining background basically and wanting to understand the technology more and the use of the technology, it was just natural that I was going to study data analytics. Data analytics and big data was really the hype around that time.
Andreas Deptolla (04:23.379)
I'm still on the app.
Stella (04:47.086)
And I converted completely, I became a qualitative researcher, which means that I mainly use interviews and ethnographic observations to understand what people do, how they work, what they do with the technology. And I spent two years in a telecommunications organization, in the business to business sales, trying to understand, they had already introduced this data analytics software, or Modela basically.
And the data scientists were very proud. That rings a bell to my prior experiences. However, the account managers didn't want to use it. So I spent two years studying why they didn't want to use it, what they were doing with it, and eventually how the thing got implemented in a kind of like forceful approach, let's say. And that was the main study of my PhD.
In the meantime, AI, we started hearing about AI becoming the next hype. And that's how I ended up at the University of Cambridge, following my PhD as an assistant professor of information systems and continuing research on algorithmic technologies and more advanced AI technologies. And yeah, that's my long journey.
Andreas Deptolla (06:14.195)
I want to go back to that moment you described. think it was during the time in Utrecht where during the internship you developed the prototype, you were proud of it, but then it wasn't really adapted. So to follow up question, now in hindsight you're always smarter. Why wasn't it adapted and was there maybe...
later on, did something change where people then actually use it? was it not accurate enough or was it not the right amount of change management or training? Like, what are your conclusions here?
Stella (06:51.854)
I think, I mean, technically it was good enough. Of course it could definitely improve. It was just a prototype, right? But I think the challenge was one, indeed the people didn't have necessarily training around data mining, but at the same time they didn't necessarily have the resources to prioritize this as the thing. And I think one of the reasons, you know, if I go back,
Andreas Deptolla (06:58.003)
Mm-hmm.
Stella (07:19.574)
Obviously, I'm putting assumptions now. I didn't keep touch necessarily with that firm in subsequent years. However, what makes sense from looking at it as an academic, having looked at other cases, cetera, I think that they didn't invest enough resources because it comes down to, things are working well the way in which they're working. And then...
Starting to cluster your customers, for example, based on their users, that's massive transformation later on of your marketing functions, right? So it requires a whole reorganization. It requires training in changing the mindset across a lot of different professionals and specializations. So I think that this is the deeper reason why that was not... And I have seen that in so many different contexts since then.
Andreas Deptolla (08:13.917)
Change is always hard. Now you described your career and also moving out of your country and having international exposure and whatnot. If you now reflect on your career so far and you could kind of go back to the beginning of your career, is there something that like...
you would tell yourself to do differently, right? Or what advice would you give yourself?
Stella (08:49.304)
Sure, that's very interesting question. think one advice, which is an advice, it's a piece of advice that I also give to my students nowadays, is I did lots of moves. I was very versatile and flexible, but I always did that feeling that I was taking a risk, that I was changing. And I think I would just go and tell myself, relax.
what you're doing and getting exposed to different ways of thinking, to different disciplines is actually only helping you. So it's not necessarily a matter of doing it differently, but doing it with more confidence. And I think maybe one thing that I would have appreciated more would have been taking more time and read philosophy. Because I think when you think about it, it always comes down to philosophy. And I think that would have been helpful for
Andreas Deptolla (09:38.195)
Mm.
Andreas Deptolla (09:42.449)
Hmm.
Stella (09:46.616)
for me as an academic, also for, think philosophy is helpful for anyone basically for how you approach and understand things and getting that frame of mind, that way of thinking about the people in your organization as well as about the technology.
Andreas Deptolla (10:05.811)
And was there like a certain research paper or project or anything like involvement where you said like, okay, this was really a career booster for me from an academic perspective. And maybe I asked a second question, was that on the flip side, was there maybe something that like was a failure, right? And maybe that failure now looking back at it had actually resulted into interesting learnings and whatnot.
Stella (10:26.957)
Mm-hmm.
Stella (10:34.318)
So something that I think one thing that was definitely a booster, well, not necessarily booster because I worked really hard for it. The paper that I published reporting on the findings from my PhD study of the telecommunications organization, it was published in the Journal of Organization Science in 2021 together with my co-authors.
and Farage and Marlene Hausmann and Hans Berens, I think that paper was really fundamental. I'm very proud of that paper, not only because I worked hard for it, but I think it really brought a different perspective to the field of management about how we think about algorithmic technologies, given that the typical narrative is always around bringing in efficiencies, increasing productivity. And what we analyze in that paper is how it actually
deeply challenges and transforms the nature of knowledge, how we know, how we produce knowledge, and that then has a significant effect in the work practices. So that was a new perspective on understanding those technologies, which has shaped a lot other people's research in our field. So aside from the fact that it was published in a top journal, so you could see that as a career boost.
And as well as that it's been very well received. It is, I think, one of the papers that people mainly know me for in my field. I think it has had a deep impact as well. the case that we report on, I have also converted into a case study, which I actually use in my MBA classes, in my executive education classes as well. So executives always relate to that story.
Andreas Deptolla (12:20.849)
Mm-hmm.
Stella (12:30.764)
In a way, I it as impactful as well because it offers food for thought for executives and leaders who are currently thinking of how to implement such technologies in their organizations.
Andreas Deptolla (12:41.931)
Now, if somebody is interested in the case study, maybe as a little teaser, what are kind of like the key one or two learnings, right? The key takeaways from a perspective of applying this, right, to day-to-day management.
Stella (13:00.11)
I'm gonna fast forward because obviously you can say I could be talking about this paper for very long time. I think a very important learning is, as I said, when you think about implementing such a technology, I think it requires the managers to go deeper than just assuming there will be resistance, but there's always resistance with any kind of technological change. But I think it requires managers to really go deep and understand
Andreas Deptolla (13:20.851)
Mm-hmm, of course.
Stella (13:28.248)
how such a technology is gonna shape knowledge and start kind of like unpacking what are the practices that we use for producing knowledge that we are fine with assigning to the AI tool, for example. What are the practices that actually we think, no, we still need the human skills. These parts of the work.
cannot be automated, they cannot be performed by a machine, it is not helpful. In my case, when it came to sales, managing the relationship with a business client, right? Of course, there's a lot that we've already automated in commercial sales, but when it comes to business sales, developing a trustworthy relationship can go a long way because you gain that intimate knowledge from your client that you cannot find in data. That knowledge is just not data-fied. It is a matter of...
Andreas Deptolla (14:09.713)
Mm-hmm.
Andreas Deptolla (14:23.635)
Mm-hmm.
Stella (14:25.49)
get developing that close relationship where your client is going to trust you and say, you know, I think I'm going to do this strategic step in the in the next couple of years, which is then translated into business opportunities, portfolios, etc. automatically. And this is something that, you know, this is just an example of something that could be lost if you only think about efficiency. And the other thing that is massively important is when you introduce such technologies.
You don't just introduce the tool, you introduce a set of technologists who enter the workplace together with the tool. You will bring in data scientists, will bring in developers, and those people really need to actively engage and collaborate with the business managers. And that's something that we often overlook. It was definitely overlooked in my case. And you need to manage that cross-boundary collaboration, basically. You need to enable that cross-boundary collaboration.
And that requires investment, not necessarily financial investment, but also investment in terms of, those people need to spend time together. It's not going to work with just a, I don't know, a monthly kickoff meeting. They actually need to be collocated. They need to understand each other. The data scientists need to understand the work functions that they are aimed to augment, let's say, to support with their tools. They need to understand
what matters, is the critical information. They need to understand what is at stake basically. And at the same time, the business managers, they need to start appreciating that different mindset. And that cannot happen overnight. It requires training, but it really requires building that deep relationship and that's trustworthy relationship between the data scientists and the business managers.
Andreas Deptolla (16:17.273)
It's funny, a recurring scene with all of these recordings so far. Of course, there's technology and AI and whatnot, but at the end of the day, it's us as people. We got to connect. We got to make the decisions. What you're describing, this framework of how to essentially set it up for more success.
Are there certain known companies or case studies where people have done it successfully that people could look up and get some inspiration from?
Stella (16:52.879)
that is a good question.
I think it depends, I think you would have to figure out a case that is comparative, right? So when you look at the more tech savvy organizations, I think it depends on the work domain and it depends on how tech savvy, what is the level of the digital transformation and the maturity, let's say, of the organization overall, right?
Andreas Deptolla (17:06.577)
Mm-hmm.
Stella (17:26.247)
I'm not necessarily sure I can give you an example of an organization that was way far back in the digital transformation journey and they made it all the way through. I can actually, can give a few examples. I'm not sure if I can say names because some of the cases that come to mind have been clients. One is, for example, a CPG company that I can think of.
where they, I think they, what they did very successfully was one, developing a culture of experimentation, really going into first trying out small experiments to figure out what works and what doesn't work before fully changing their whole function. Another, and the other thing that I really enjoy is that actually they have
they really try to promote innovation. So for example, one of the managers was telling me, we cannot meet on the Friday afternoon. We don't do any meetings at that time. This is the time that each one of us takes to think, read, and basically figure out what they want to do. Which is obviously, that is translated in resources, basically, right? This time is important, but that CPG firm shows to prioritize it.
Andreas Deptolla (18:33.523)
Mmm.
Andreas Deptolla (18:45.715)
Of course.
So similar to what Google did in the early days, I don't know if they're still doing it, but they said, like, everybody can use 20 % of their time to work on their own projects, right? And commercial viable product came out of that.
Stella (19:00.66)
Exactly. And you know, think that goes. Yeah, and actually that goes a long way because then you you empower people, I think a lot of times. I mean, you do need a top down approach to digital transformation. Absolutely. You need to have a vision. The leaders need to set up vision so that the rest of the organization can follow. at the same time, it's quite important to allow people to experiment because they they know their work best. They know what works best.
rather than having a new tool or a new way of working land on their desk, they need in a way to explore it. And they may come up with interesting use cases, interesting solutions that you wouldn't have thought. So you need to give them the time, but also you need to empower them to allow them to feel that they have authority over saying how things can be done. obviously that depends on the level, but
at the very least, can you engage your employees on the operational level, on the training of the tools? That is already a step towards empowerment, or allowing them to have the agency to feedback and say, hey, this AI tool is not necessarily working. I have my doubts whether this recommendation is the correct one. Giving them that agency at the very least is going to go a long way to enable transformation. Yeah.
Andreas Deptolla (20:26.127)
I'm sure that also has a big impact on job satisfaction, right? If you know, hey, here's a clear vision, right? We can go behind that and the employees have some authority ownership over it, right? That makes a lot of sense. Now, let me maybe ask you a question from a slightly different perspective. So now in your teaching, right? Or maybe even in your personal life, like...
Stella (20:41.869)
Yeah.
Andreas Deptolla (20:55.443)
How are you using AI? not so much in, hey, Cambridge is spending thousands of dollars on Nvidia ships or whatever that is for computing power. in terms of products that are $25 and below, are there certain tools, applications that you personally use?
Stella (21:22.734)
So it's interesting, I'm a tech enthusiast. I don't have much time to use a lot of tools. think from the latest tools, the only tool for which I paid the subscription was Judge GPT. And right now I'm not paying a subscription because it is actually provided by the judge, by the Judge Business School specifically. I have...
Andreas Deptolla (21:39.1)
Mm-hmm.
Andreas Deptolla (21:46.383)
All right.
Stella (21:52.268)
I have tried other tools as well. I tried Elicit because in theory it sounded really fascinating. Elicit is more of an academic tool using GenAI aimed to support you in literature review. I think at least at the time that I tried it, they didn't have access to the major databases. So was very difficult to test whether it would really work. My co-authors are using
Andreas Deptolla (21:54.813)
Mm-hmm.
Andreas Deptolla (22:13.906)
Hmm.
Stella (22:19.754)
the anthropic model, is it called cloud? Cloud, right? And notebook LM as well. So I haven't really had to use it myself, but my co-authors say a lot of times when we say, okay, we need to do this fast. So they give a prompt. And I've been looking at those tools, I think in terms of research, I'm sorry to say, I don't find them helpful.
Andreas Deptolla (22:21.5)
Mm-hmm.
Andreas Deptolla (22:47.783)
Yeah, and that's where we are right now, right?
Stella (22:48.97)
Unless you really... Yeah, think the best thing in which they have helped us is saving us little bit of search term, term search basically. So when we really know we need this and this and this, and then we can give the prompt to the tool and say, I'm giving you all this data, can you go and search and figure out where these things are mentioned basically?
Andreas Deptolla (23:00.509)
Mm-hmm.
Stella (23:16.814)
But it's gotten us already to a stage where we know exactly what we're looking for, and it's saving us a little bit of time. And then we still have to go and read the whole thing and do the interpretation, because we do interpretative research, basically. I've tried, so I use Atlas.ti, which is a qualitative analysis software for qualitative research. they have GenAI. I've experimented with it.
Andreas Deptolla (23:21.267)
Sure.
Stella (23:43.958)
sorry to say I never found it helpful, but I think it really has to do with the nature of the research. Where I do find those tools helpful, especially ChargPT, the paid version, which I have used a lot, definitely, it saves me lot of time when I have to do very, very mundane tasks. Like for example, last academic year I was serving as chair of examiners on an MPhil program. At the end of the year I had to write my report.
Andreas Deptolla (24:02.864)
Hmm.
Stella (24:12.184)
This report looks exactly the same every year. basically what I did was I gave the tool last year's report and I said, okay, this is my last year's report. These are the things that are important from what emerged from this year. Can you write me up the report? Yes, that did it in one minute. I read it, it looked perfect. Yes. So for very mundane stuff, yes. For marking, no.
Andreas Deptolla (24:32.541)
Totally, yeah. And then you have to polish it, but yeah.
Mm-hmm.
Stella (24:42.094)
It really doesn't work. I don't know. Yeah, no. It's only for very basic admin stuff. Actually, where I found it helpful was on personal use. I asked it to look at specific websites with information. I was looking for something very admin related. And I said, OK, tell me the rules. Tell me.
Andreas Deptolla (25:09.468)
Hmm.
Stella (25:09.544)
what is in this and that, and it was very good at summarizing and kind of like pinpointing me to the right direction. So yeah, I think I'm not the most enthusiastic, but at the same time, when you think about automation, saving time in very basic, simple tasks, yes, absolutely, those tools are very, effective, and they can have an impact on productivity.
Andreas Deptolla (25:32.113)
That's about that. And what about your students? What do you see there? Are they now writing the essays with the GDP? Can you tell? How has the academic landscape changed?
Stella (25:44.918)
Yeah, some of them do. We can tell. We can tell. I think the challenge is it's difficult to prove it. It's interesting. So yeah, so we can tell. And we have failed people, not because they used Chudge GPT, but because their paper was not good enough, basically, even though they used Chudge GPT. There wasn't any depth.
Andreas Deptolla (25:53.031)
Mm-hmm.
Andreas Deptolla (26:06.003)
Mm-hmm.
Stella (26:12.482)
they hadn't figured out that they actually had to combine what they were asked to do with their own thinking, with their own experience. so we could tell that. Very interestingly, when a paper is too well written in terms of grammar, that can be a sign. honestly, this...
Andreas Deptolla (26:18.835)
Hmm.
Andreas Deptolla (26:32.347)
I was too well written.
Stella (26:37.59)
And it's interesting, I hope I'm not exposing none of my students' laughs now, but from the module I just finished marking last week, I had the feeling that nobody had used it. They may have used it for search, and I actually encouraged them. I said, you know, you could use it for brainstorming. Of course, in theory, they have used it, they have to report what prompts they used, et cetera.
Andreas Deptolla (26:42.823)
Hahaha
Andreas Deptolla (26:49.329)
Okay.
Yeah, ideation or something like that, right?
Andreas Deptolla (27:02.643)
Mm-hmm.
Stella (27:05.466)
But you know, practically, yes, it can save you time. you know, imagine like you don't know where to start from. It can give you some points. But of course, then you have to do the due diligence. One, like is this thing accurate? Two, it's not just enough of giving you a few facts, at least for the papers that students have to write in Cambridge. You actually need to interpret what is happening. And then...
Andreas Deptolla (27:13.191)
Like a structure.
Stella (27:31.24)
you need to master the readings that we have assigned you as a student to read and you know those things we can easily tell if someone has used it or not. The big question for me is are they able to pass the class or not if they have only used the tool? I'm not sure I have the answer to that. I think maybe someone could pass.
Andreas Deptolla (27:50.639)
And would they pass an oral exam now? mean, now you don't have that crux anymore. But I think you said something earlier that you would actually encourage them to use it early on, maybe for some ideation or structuring or whatnot, framing.
Which makes sense to me, right? Because at the end of the day, like, you know, most of these of your students, specifically on the MBA word, you want to prepare them for the real life, right? And, you know, obviously people will use AI and they got to be able to use it in a meaningful way. If we now look at like kind of outside of academia, right? You and I talked a little bit about knowledge workers, right? How is it affecting them?
And maybe with knowledge workers, to kind of like give some examples, right, we really think about like the traditional accountant, the lawyer, right? In your view, how is AI going to reshape their profession? And maybe now, but then also if you just look at like, you know, the near future, next three years or so.
Stella (29:04.204)
A very interesting question. So I think in professional services specifically, so you mentioned lawyers and accountants, think technology a little bit more broadly, even less advanced technology, like for example, let's say robotic process automation, right, which may be quite rule-based, it already has massive impact on what they do. So I started studying lawyers before...
Andreas Deptolla (29:13.149)
Mm-hmm.
Stella (29:33.454)
before the pandemic. Of course, AI was already a hype, but most of the tools were not using AI somewhere already. But what is interesting is that even basic natural language processing tools can have a massive impact on lawyers' work because it's not difficult to automate routine types of tasks that they would do and increase their productivity significantly. So lot of tools actually can...
automate several tasks that paralegals and junior lawyers would have to do. And I think there's similar tools in accounting as well. If you look at, so the one aspect is the automation basically, right? Which could happen with more basic, it could happen obviously with more advanced AI tools. The other aspect, which is very interesting is the augmentation of several tasks as well, where you don't necessarily have the tool fully take over. So the automation, when I talk about automation for me, like the tool is carrying out the task period.
Andreas Deptolla (30:28.051)
Hmm.
Stella (30:32.418)
But you can also have lots of cases of augmentation where the task, in the task you have the human carrying out the task but being augmented by technology. And that's where AI is bringing in new types of impacts as well because in professional services, for example, it allows people to access information and insights that they wouldn't have had access to. Like imagine, for example, a law firm, right?
They're in the data rooms, they have access to so much data, so much text that can now actually be analyzed and it can give, it can support litigation, for example, with finding evidence for cases. It can support due diligence tasks, assessing contracts, there's any kind of area basically.
And I think, there's interesting opportunities, productivity definitely, one. Two opportunities, I think, for providing new types of services actually to the clients, right? And we see that, like we see with, so for example, one of the cases that I have found very, very interesting is a case by my colleague Karim Lakhani on Deloitte Pixel. So Deloitte,
Andreas Deptolla (31:41.907)
Hmm.
Stella (31:57.312)
So the case is basically starting from the pressure that consultancy firms and advisory firms are receiving from their clients. The clients are saying, well, you have access to so much data, I want those new services. So the clients are expecting those new services. So there's a lot of opportunities for portfolio diversification as well, and for really offering that more
Andreas Deptolla (32:12.604)
Hmm.
Stella (32:25.934)
boutique, let's say, type of service to your clients. Now when it comes to the work, however, right, because initially your question was about the work, then there's interesting opportunities and challenges. think what we see is that actually from juniors who are mainly the ones who are affected, actually there's not much resistance there. Juniors are very enthusiastic, right?
Andreas Deptolla (32:50.583)
Mm-hmm.
Stella (32:53.77)
And I think there's an element of one, them not wanting to do the work. The other thing is you have to place it also in the context of the juniors of today are Gen Z. And there's a lot that has been said about Generation Z of sort of wanting a different work-life balance and expecting to work in different way, expecting that yes, that we can use technology, right?
So juniors are definitely excited because they feel like, we're going to be more productive. We don't need to spend 18 hours over a day for the next three years being locked in the data room. That's how a junior lawyer used to work. And they would have to do exactly the same task again and again and again. And they feel like, well, I'm not learning anything through that. Now, there comes the question, however. And so there is resistance.
from more senior people in the firm, from the partners. And the partners say, well, no, you have to do that because that's how you learn the law, for example. That's how you get trained. You need to those mundane tasks, which may feel boring. But hey, that's absolutely necessary to develop your expertise. And so I see kind of like a contestation of like, what does it mean to be a legal expert? What does it mean to be?
Andreas Deptolla (33:56.199)
Mmm.
Stella (34:17.878)
an accounting expert, how do you develop that expertise when the routine tasks are taken over by tools? Is a junior lawyer ready to go and develop an argumentation for court? Is a junior lawyer ready to go and have a client meeting? Those questions about the nature of the work changing and the development of the expertise, they're interesting questions which we need to answer as.
we implement those tools in professional services.
Andreas Deptolla (34:50.195)
And if you would kind of like classify these different cases, it sounds to me that like what you mentioned is like maybe more entry level work, right? This could potentially be to craft like a standard.
non-disclosure agreement or a simple contract, what not. It seems like AI can probably do a decent job on that, right? If you give it the right input parameters and somebody needs to look over that. You also kind of like painted that picture in my head, at least that kind of like Hollywood movie, right? Where like the junior lawyers would have to go through box and boxes of material, right? And somebody could envision that as well, right? Everything is now scanned, digitalized,
Now I can find the right documents. But then you mentioned that from that, there could also be now new service offerings that the law firm could provide. Just to maybe brand some on that, are you thinking like, now I have all of that data in a structured way, and I can read it, and now I can maybe find certain nuggets that are not even related?
to the business case, like, hey, did you know that you also own this IP, right? Or did you know there are certain other things? Is that where you would go with that? Or did you have something else in mind with like kind of other ways to monetize the service?
Stella (36:21.39)
think there's different ways. So yeah, it could be totally new service offerings, like new advisory services, for example. So yeah, it could be from we have data about you, or it could be, well, from the data that we have from our clients combined with data that we have from the environment, we could actually be giving you more strategic decision support, like what market, for example, you could be moving. So imagine.
Andreas Deptolla (36:26.056)
Mm-hmm.
Stella (36:49.376)
an accounting firm developing new types of services that may not be directly related to accounting, for example, but which, however, they are in a position of being able to offer, obviously with the right talent acquisition, development of a new capability, et cetera, could be just improving the current value proposition. So let's say, well, we already do, let's say, for example, risk assessment services, but now we can do more sophisticated.
risk assessment, where we take into consideration many more factors, details, et cetera. Of course, we could be talking about scalability as well. But of course, then you actually have to look into the structure of the market as well. And that all sounds exciting in theory, that there's lots of opportunities for what I would call basically business model innovation and creating value in new ways.
Andreas Deptolla (37:21.779)
Hmm.
Stella (37:44.322)
But the question is, can everybody do that? I'm not sure if they can. So can a small or even medium firm fine tune and have their own model based on their own data? At the moment, at least, it looks like it would be a very pricey endeavor. At the same time, what about the big law?
Do they have the versatility to do all that change? So I'm not sure how the market is going to restructure and who is going to be able to, you know, to really innovate. And also who has the stomach to do that as well, right? Because business model innovation is not easy. Like the business model is really ingrained in the firm's DNA. It's ingrained in the culture, in the practices. It's not easy to take your whole firm, especially a larger firm.
and change the complete mindset.
Andreas Deptolla (38:49.349)
Yeah, it seems to be point, it seems to be right for disruption, right? Because I mean, like I haven't seen much. I mean, it's typically money for time, right? Billable hour, right? That's kind of how it works in most cases, right? And the questions are the incentives really of the firms and the clients truly aligned with that, right? Have you seen any other interesting
Yeah, new business models that are on the horizon, right? That could be utilized, you know, whether by accounting or by law firms, right? That would maybe take advantage of some of the opportunities.
Stella (39:32.28)
So I think there's a couple of things. is, so yeah, you mentioned the billable hour, right, which is very interesting. So I remember last year we had the top accounting firm doing some executive education with them. And there comes information from the head of digital that a partner had hired the junior.
to carry out a task, they had estimated that it would take about two days and the junior ended up writing a script and doing it in 10 minutes. And obviously, you know, that's, you know, we can say a lot about how tech savvy that partner was, et cetera, but practically speaking, right, the question is, okay, you know, what do we do in the rest, I don't know, like 15 hours and 15 minutes, basically, of those two days worth?
Andreas Deptolla (40:06.205)
Good for him.
Mm. Mm.
Andreas Deptolla (40:16.189)
Mm-hmm.
Stella (40:22.702)
to charge the client if you go by the billable hour. So I think that it requires changes in the revenue model. I think there's interesting endeavors, innovation endeavors happening by big firms where they actually, can afford to have their own incubators, so they're supporting lots of startups. They acquire some of those startups within those incubators to integrate them in their businesses. So we could be seeing kind of like the mass market
part of services being quite automated and relying on platforms. I'm not sure on that part of the market, I'm not sure where the disruption is going to come from. I don't think anybody knows at the moment. Everyone is trying from different perspectives, from big tech to low firms, from small low tech startups. We don't know who's going to be the disruptor there. And I think that where, most likely, what
the type of model that is going to be successful will be the one where people still rely on relationships who are maybe in more boutique type of firms, Where the, or firms that are really specialized in a specific area and where you would expect the relationship with the clients to still matter massively, where the clients actually need that more nuanced understanding that a human can offer. A human augmented.
by AI insights, but still where you would need those specific professionals and you wouldn't easily just turn to a competitor, for example.
Andreas Deptolla (41:48.903)
Mm-hmm.
Andreas Deptolla (41:59.975)
Yeah, so maybe a combination of technology, right? Where the technology reads all of your receipts and pay slips and whatnot, puts together your tax returns, but then like, you still want that person to look over it, right? You still want to have that personal relationship and the oversight.
Stella (42:35.384)
Sorry. Yeah.
Andreas Deptolla (42:44.925)
Perfect, yeah, so we spoke about the...
how likely skilled service workers like lawyers, like accountants and others will change. And it's certainly not easy to disrupt these industries, right? We have seen this in the United States with real estate brokers, right? People have discussed for decades, right? You know, and try different models. So, what would be interesting where that lands with technology.
thing that like, you know, you and I touched base on the other days in terms of if you think about like changes and disruption is Wikipedia in the sense that like Wikipedia was kind of like the go to place, right? If I want to learn about a certain topic, right? And was all user generated, right? But now AI is generating so much of the of the content, right? How is that affecting it?
Stella (43:45.218)
Thanks a lot for asking that. And actually you're taking me little bit back to what aspects we were discussing earlier. Of course, as you've seen so far, everything that I do research-wise looking at AI is around knowledge and knowing. you're mentioning Wikipedia. I have been doing research with a couple of really wonderful colleagues on Wikipedia specifically because we think that when
when you look at GEN.AI, this is a very interesting case of how an organization and actually how their, what I call the regime of knowing, the way in which knowledge gets produced and used is gonna change, given the changes on GEN.AI. Yeah, so that's a very interesting case. I think with Wikipedia specifically,
What is happening is obviously yes, Wikipedia is web, it's available on the web. produces, it promotes this community-based approach, right, towards producing knowledge. So even though they have been quite exavid, they have been using bots for years, I think that the...
Andreas Deptolla (44:58.515)
Mm-hmm.
Stella (45:14.582)
Yeah, you know, is actually changing now is that GenAI is kind of like bringing in different threats and challenges from sort of the for most sorts of different directions. one is it's obviously, know, you it's been affecting, I think, production of information.
Andreas Deptolla (45:39.803)
Mm-hmm.
Stella (45:40.458)
in multiple ways, right? So one, obviously, it could be used as a tool by the volunteers themselves. Two, yeah, exactly. And at the same time, there's a lot of web content that is currently available that has been GNI generated. And I think that brings several challenges because there's two aspects, right? Obviously, the important
Andreas Deptolla (45:45.757)
Sure. Easier to create content, yeah.
Stella (46:07.054)
The most important aspect for the Wikipedia movement is open knowledge, right? This is what they care about. But at the same time, what they also really care about is not just open knowledge, but actually providing valid information. So they want to always verify the sources. You're not allowed, for example, to report on any original research on Wikipedia because it needs to be verified, right? So the verifiability aspect, the aspect of
Andreas Deptolla (46:13.107)
Mm-hmm.
Stella (46:36.302)
of attribution and ensuring the quality of the information that you're reporting on, this is massively challenged. it is a big challenge because it is difficult for anybody to assess. I told you earlier about my students. I'm not sure if they have used GenAI. There's always the question mark. And actually, there's been cases like one of our interviewees from the Wikimedia Foundation.
was mentioning, for example, that, well, there was this case with Sports Illustrated that came on the news, I think a year ago or so, that there had been articles that had been GNI generated. so GNI is too original for Wikipedia, right? There's always a part that maybe you don't know how it is produced. It is generated. It's too original. So it was a challenge for them because until then, Sports Illustrated was seen as
a very reliable source and then it brings a question like, well, can we trust it? And the other aspect which is very, very important for Wikipedia is of course the consumption of information, right? The consumption of web content. And that's an area that is increasingly challenging because obviously we've seen a shift in how users consume information more broadly already through the use of social media, et cetera. But the interesting thing is that
now it's becoming even more challenging because there's no attribution specifically, right? Maybe there's a few links to sources, but LLMs are not built with attribution in mind, right? That's the algorithm basically behind it. And that is a challenge because then without attribution, there's much less movement towards the Wikipedia.
page. if users don't end up in the Wikipedia page, how are they going to support it? Either through voluntary work or through funds. So it could have an impact on the Wikipedia community over time, basically.
Andreas Deptolla (48:47.249)
I wonder if what now seems to be a threat for Wikipedia could actually be an opportunity. We're seeing now the search traffic is changing. It's going away from Wikipedia, it's going away from Google to TGP and other things. It's slowly, that's at least the direction.
you mentioned the importance of sources, right? And trust. I'm wondering if Wikipedia could do something where I said like, Hey, what, what, what is really unique about us is our process to verify our community, right? So this is kind of like the gold standard here. And now you create an app that sits on top of it. Maybe, you know, you got to pay five bucks per month or something like that. Right. But like, now you can ask those questions and you get.
all the sources, right? It seems like to perplex it here, some of those things, right? Like that might not be the solution here, right? Exactly. But I think, you know, it sounds to me that like Wikipedia would have to adjust, right? And just as a new, new reality.
Stella (50:05.294)
Absolutely. And I think that's actually the fascinating thing with Wikipedia. Of course, we all know Wikipedia sort of like being known as the disrupter of the Britannica. They don't want to be the new Britannica. They want to embrace the change and they're trying to figure it out how to adapt. I think what is very fascinating with Wikipedia is again coming back to what we were saying earlier about
Andreas Deptolla (50:17.971)
Mm-hmm.
Andreas Deptolla (50:22.64)
Exactly.
Stella (50:35.566)
You need to encourage everyone to experiment, And that is the beauty of the community approach, basically. Anybody from the community can try out different tools. A lot of members and especially affiliate groups of the movement, they collaborate with academics, for example, right? So they are exploring a lot of different ways to...
to figure out how to embrace those technological developments in ways that are still aligned with the core values that they have about knowledge. So hopefully they will remain relevant. I think they need to remain relevant because they are the providers of truth. They are the providers of true facts. I think the...
the chief operating officer of the foundation, actually said it really nicely, basically, if they want to, if the big tech want to have models that work and that produce good facts and good information, they need Wikipedia at the very least to train those models. So hopefully we're gonna find, they will find a model that works.
Andreas Deptolla (51:59.251)
That is changing, right? read the other day that a lot of the models now, I Wikipedia is one element, right? Of course, the bots are scrolling the entire internet, but more and more weight is giving to Reddit.
And that's interesting, right? Because that's all user generated content, right? Not necessarily peer reviewed. So yeah, certainly interesting, like, you know, what goes into those models, right? How will this change? Stella, unfortunately, we are getting to the end of our time here. think we could have probably filled another podcast here together.
I want to make sure we cover a couple of other things, like specifically like for your students, right? You know, whether it's now the MBA level or different levels, right? Like, there's a lot of advice out there of what to do, what not to do, right? What are maybe some misconceptions about AI and how it will influence a career and whatnot that you find among students and that might
not resonate with you, you might have a different opinion there.
Stella (53:12.93)
So I think, you know, definitely something that at least obviously I see as a misconception is, you know, thinking that they can rely on the technology to do the thinking for them. And, you know, I think that they have to figure out they cannot, you cannot turn the blind eye to the technology, of course, we're going to need to be more productive. So they're doing very well, but they're experimenting with it.
Andreas Deptolla (53:26.579)
Mm-hmm.
Stella (53:42.082)
But I think they must not underestimate the work that they have to do in terms of being able to understand, how should I say, you know, in a way the world more broadly, right? You need, the fact that you have a technology that can summarize a book for you doesn't necessarily mean that you should stop reading books and only read the summaries. Because it's, don't just read the book to get a bunch of facts.
Andreas Deptolla (53:54.791)
The fondage, yeah.
Stella (54:09.826)
you read the book because it changes your perspective. So the way in which something is written, I'm not talking about all books, right? But there's a lot of books that you need to read because that will shape your perspective, that will shape your worldview, how you think. You still need to develop, you need to learn how to learn, you need to learn how to think, how to approach problems, even if you end up bringing in the technology.
Andreas Deptolla (54:22.535)
Mm-mm.
Stella (54:39.274)
in the resolution process of the problem, but being able to understand the problem on the first place and being able to resolve it, right? So I think in a way, yeah, misconception is that technology can automate, so we don't need to think anymore. think that's something that they need to address.
And I think that we as a humanity actually need to figure out how to address because it's not just the MBA students. I think it's something we need to take into consideration across all education, across our workers. As I said earlier, you need to go back and look at the work and identify what are the aspects where we need humans to think, we need humans to feel, right? And make sure that we protect those and make sure that actually people are
trained to cope with those situations because I think if you rely too much on the technology, you might not necessarily be able to cope with those outliers or identify how to approach problems, cetera.
Andreas Deptolla (55:46.003)
It's interesting that you mentioned the books and having summaries or not. mean, this is obviously nuanced. Certain books are certainly worth reading every single page. I always felt it was business books. Most of them should have been a blog post. And then you add 200 pages of examples to it.
Stella (56:06.914)
I was not talking about business books.
Stella (56:14.094)
Yeah.
Andreas Deptolla (56:14.707)
I think this is actually a great application of AI. like, know, summarize that for me if I'm now really interested in it, like, okay, I spent the hours to read this thing.
Stella (56:26.478)
Absolutely. I'm not, no, I was not referring to specific management types of books that you would quite likely find at the airport, for example. Yes. And actually a lot of which I fear that they may now be largely written by GNI potentially, or they could be in the future. No, but for example, my...
Andreas Deptolla (56:34.252)
Yeah
Andreas Deptolla (56:44.381)
I've sent down.
Stella (56:52.598)
My PhD advisor, Marlene Hausman, she was a student of James March, one of the biggest organizational theorists. And in his classes, he would always assign the Don Quixote book by Thurvandis. But he saw that as a book to learn about leadership, for example. So that's what I mean in terms of developing an approach to problems.
Andreas Deptolla (57:04.419)
Mm-hmm, that's a difficult read.
Andreas Deptolla (57:10.323)
Mm-hmm.
Stella (57:22.102)
It's not a matter of, I don't think, you cannot move forward with a technocratic approach where you're like, okay, I this set of skills and, you know, these set of guidelines to develop those skills or to address those problems. We need that broader frame of mind. So yes, thanks for, you know, for allowing me to clarify. It's not all types of books.
But also equally, when you read the case study, example, a lot of times the devil is in the details, right? So there is a reason why we spent 16 pages to write a case study because we want the student to actually really grasp and understand the whole problem, understand all the contextual environments, the contextual factors that shape the situation that the firm is faced with. So I cannot see how a student would actually learn
Andreas Deptolla (57:54.643)
Mm-hmm.
Stella (58:18.136)
how to approach that themselves. really, you know, when, because in the future they're going to be in front of a client, they have to be on their feet. They cannot just rely on the GEN.AI tool to give the response, right? So.
Andreas Deptolla (58:29.683)
Yeah, you have to understand the fundamentals, right? mean, like the same now goes for programming, right? Like there's an argument to be set, like, know, students learn how to like, you know, write in Java or C++ or whatnot?
It seems logically for me that like, at the very least you get to understand the structures and how it's done in order to influence the AI, right? And essentially lean it in the right avenue.
Stella (59:02.688)
Absolutely. mean, I think it's the principles. Again, it's the way of thinking that you need to embrace. And once you master the way of thinking, the way of approaching the problem, I think then it becomes then, you you can find lots of different ways. Honestly, I, you know, I did I learned C, C++, Java, when I did in computer engineering, I don't remember how to write anything. But I think but I'm pretty sure, you know, I have that that understanding like that.
higher level understanding, which I think is very, very helpful to resolve problems. Now that said, did recently hear, so my husband who's also an academic, he was recently at a talk in the master's on data science. And someone from one of the big tech working on GEN.AI was presenting and was saying that actually they
Andreas Deptolla (59:31.527)
Mm-hmm. Very important.
Stella (01:00:00.558)
There is this book on algorithms, a very famous book. They tried to train the model on those basic algorithms expecting that the model would actually be able to start understanding the reasoning and the logic behind those very important, kind of like fundamental approaches to problems. But I'm not aware of the results of that attempt, to be honest.
Andreas Deptolla (01:00:26.077)
We did something, I mean, probably like a lot simpler. We put like a little problem on the whiteboard and we took a picture and uploaded that into TGP and just like, just solve it with code and it was perfect. Right, so, you know, I think where we probably are is like that AI can help with these simpler things, right? Whether it's, you know, we talked about contracts right now, maybe.
certain simpler things on the coding side, whether bug fixing or security reuse or not. like, yeah, we still need to understand the fundamentals in order to create something really meaningful.
Stella (01:01:09.784)
Absolutely, yes.
Andreas Deptolla (01:01:11.453)
So my last question, if you could, if you have the magic wand, right, who would you invite and would like to see here on the podcast next?
Stella (01:01:25.518)
So, okay, if I had a magic wand, I think I would invite together Herbert Simon and Hubert Dreyfus, ask them to reflect on the current developments in AI and kind of like they, you know, revive their debate. They had a long debate around cognition.
Andreas Deptolla (01:01:41.811)
Mmm.
Stella (01:01:52.302)
human cognition versus AI, can machines think, what can machines do or cannot do? So I would be very curious to hear them talk nowadays. actually, yes, so I will be very curious about that. Now, maybe my magic wand was not that effective if I were to invite someone alive, I would.
Andreas Deptolla (01:02:08.275)
That's a good one.
Andreas Deptolla (01:02:14.705)
Hahaha
Stella (01:02:17.9)
I would highly recommend inviting someone from the Keen Center for Digital Innovation in Amsterdam, not just because they're all really amazing friends of mine, but I think they're doing incredible work. Especially there is the AI at Work Lab that is embedded in the Keen Center who are doing all of the deep ethnographic work on data and AI. For example, one of the...
Andreas Deptolla (01:02:35.283)
Mm-hmm.
Stella (01:02:46.026)
one of the PG students, she's been looking at the use of biometric data in sports. So she would have wonderful stories to share about what biometric data can do and what it can do, but in a different way from what we would have anticipated, which is very interesting.
Or for example, another person, she's been doing a lot of research in healthcare, in the use of robots in healthcare. So she would also have incredible insights to discuss. Or the head of the group, Marlene Hausman, with a lot of experience, she could touch upon lots of different AI cases. I think I've given you enough recommendations.
Andreas Deptolla (01:03:36.059)
This is, mean, honestly what poked my personal interest the most. mean, there are so many of these devices now, like the Aura ring and the whoop and whatnot, right? In the US where it tracks all these different, you know, your sleep and whatnot. like to look at that from a more academic perspective, right? What is real? What's aphysically? I think that that would be super interesting.
Stella (01:04:07.286)
No, absolutely, yes. And I will be happy to connect you with Lorna and the rest of the team about that. think it's super interesting. As I said, there's a lot of biometric, a lot of insights that biometric data can offer. But again, it always comes down to the fact that, these are insights, but then the way in which people can use them or the way in which people are affected by the access that they have to those insights, that's a different story. So there's a lot.
Andreas Deptolla (01:04:11.251)
Thank you.
Stella (01:04:36.416)
a lot to unpack there about opportunities and challenges from biometric data in managing performance. In the case of Lorna, the PhD student that I mentioned, she's looking at obviously the very extreme case of managing high performance, but I think there's a lot we can learn from this case to reflect on using biometric data to manage employee performance overall. There's lots of wellbeing programs currently developed, especially I think in
Andreas Deptolla (01:04:44.296)
Mm-mm.
Andreas Deptolla (01:04:51.816)
Hmm.
Stella (01:05:05.78)
in US-based firms and we will be expecting, I will not be surprised if even a few years managers are looking at biometric data to assess the risk of their employees having a burnout, for example, and the moral dilemmas that arise from that are immense.
Andreas Deptolla (01:05:24.211)
Yeah, I mean, we are already seeing a movement in the US and the literature about like, know, longevity and whatnot, right? So people are clearly thinking about that, investing about that. And then there are also certain really sad case studies like in investment banking, whatnot, right? Where you can see if, you know, work-life balance is not in check, what kind of terrible consequences that can be. yeah, we'll love to have...
you know her on the show, right, and learn more.
Stella (01:05:56.374)
Absolutely. Yes. I'm to connect you with her. Yeah.
Andreas Deptolla (01:05:58.269)
Perfect. Zedda, thank you so much for your time. I apologize for going a little bit over. I know you have a busy afternoon to teach here, but yeah, this was a great pleasure.
Stella (01:06:10.03)
Likewise, the pleasure was mine. Thank you for inviting me and I really enjoyed our conversation. Thank you and best wishes for the festive break also.
Andreas Deptolla (01:06:19.603)
Thank you.