027 – Why doesn’t IDEMS simply focus?

The IDEMS Podcast
The IDEMS Podcast
027 – Why doesn't IDEMS simply focus?
/

Description

While many organisations seem to focus on a small area of expertise, IDEMS is involved in a wide range of projects and activities, in a wide variety of roles. In this episode, David talks to Santiago about the thinking behind, and advantages of, embracing diversity in how IDEMS can have impact and add value.

[00:00:00] Santiago: Hi, and welcome to the IDEMS podcast. My name is Santiago Borio, I’m an Impact Activation Fellow, and I’m here with David Stern, a founding director of IDEMS. Hi, David.

[00:00:19] David: Hi, Santiago. Looking forward to another discussion. What’s the topic?

[00:00:23] Santiago: The topic is one that I find very interesting because it’s something I struggle with a lot when trying to explain IDEMS to people who don’t know about it.

[00:00:34] David: Everyone struggles with trying to explain IDEMS to people who don’t know about it and even to the people who do know about it.

[00:00:41] Santiago: Yeah, yeah. But one of the reasons is that we do so many things. We do such a breadth of activities and projects. And there’s a question really that I keep hearing. Why don’t you just focus?

[00:00:58] David: Yep, it’s a question we all get quite a lot. In fact, it’s one of the consistent bits whenever I get mentoring advice from business leaders and so on, it’s a consistent point which comes out, and I absolutely understand the benefits of focusing.

[00:01:21] Santiago: You may not want to answer the question from this perspective, but we have a close relationship with an organisation called Statistics for Sustainable Development, Stats4SD, and that, I think, is a similar natured company.

[00:01:42] David: Well, we both emerged from the same old group in academia, in university.

[00:01:49] Santiago: And they managed to focus quite… more than us, if I’m not mistaken.

[00:01:54] David: It’s not that they managed to focus, they chose to focus. And that’s a really important point. We could focus without a problem, but we choose not to.

[00:02:06] Santiago: Why?

[00:02:09] David: Well, really if you want to understand that in full, we need to go through the principles. And we need to go through our principles as an organisation, which guide us, and one of those principles is to embrace diversity. And part of the embracing diversity is this embracing diversity in what we can do and where we can add value and how we can add value. This is deeply embedded in what we are as IDEMS.

And so it’s central to what differentiates us from organisations like Statistics for Sustainable Development who have a focus, who have a clear focus, are able to articulate it and are able to bound the things they do in that way. And IDEMS deliberately isn’t that. It’s a conscious choice that was made from day one and it’s a challenge we face on a daily basis because it does bring challenges.

But it’s not all bad, it proved extremely valuable and extremely useful when COVID hit.

[00:03:17] Santiago: Indeed.

[00:03:18] David: And this is already one of the things which has justified this choice. When COVID hit, if we had focused, then we would have been really much more negatively affected than we were. Because what actually happened is things that were relatively small parts of our work became bigger parts of our workload, and things that were big parts of our workload just went off a cliff and disappeared.

That diversity is part of what helped us during, and we weren’t the only group to have a difficult time in COVID, everybody did, but that diversity and the value of the diversity was made visible to us then. And I think more generally, there is a single point on this about stability.

[00:04:07] Santiago: Before we get into that, sorry to interrupt, for the sake of the listeners who might be joining now, can you briefly explain a bit of the breadth? Because we’re talking about doing a lot of things, but what is the range of things?

[00:04:23] David: The real challenge of the breadth is it’s multi dimensional. So, there is a breadth in terms of what we do. We build apps, we build chatbots, we analyse data, we support projects in research methods, we support groups with technology decisions, we build open source software, we offer trainings, and we offer consultancies. Those are a set of some of the things that we do.

But it’s not just diversity in what we do. It’s also in what it’s for. We work in agriculture, we work in social development if you want, we work in parenting, we work in youth, we work in education, we work in climate, we have a whole range. I would argue that the big area where we are only currently touching it a little bit but we will almost certainly expand in years to come is public health where we do a few little bits related to WASH (water, hygiene, sanitation), but there is so much more related to public health that we would probably move into.

[00:05:43] Santiago: And we also had a chat bot on sexual reproductive health.

[00:05:47] David: Absolutely. And so we have a little bit in public health, but it is an area where we know our skills are needed.

One dimension, the great diversity of application areas, we’re not unique in that. I mean, statistics for sustainable development are relatively similar. They also have a wide range of clients who work in very different areas and they bring their skills as statisticians, as data experts, to bear on a wide range of topics.

So that’s quite common for people who have strong technical skill sets to be able to work in a real diverse range of application areas and part of the value we bring to our clients is the fact that we’re exposed to these different application areas. So we get specific knowledge related to their area, but we also have knowledge related to other areas and are able to help them with innovation.

[00:06:49] Santiago: And I’m surprised that you haven’t yet mentioned the transdisciplinary principle.

[00:06:57] David: Well, I’ll come to that I’m sure and that is a really important one, but I think before we get even to that transdisciplinary nature of a lot of our work and how we work, which really relates to what we just talked about, the fact that we bring our expertise to bear, but working in and for many other disciplines and a whole variety of disciplines. And the fact that we work in transdisciplinary ways really relates to that. But I think the really unusual aspect of our diversity, in terms of the range of work we do, is the other dimension.

It’s what do we actually bring to the different teams? We have research groups that we work with, where our role is to support the research methods. We have another research group we work with, where we don’t support their research methods, they have other people who do that, statisticians who do that, whereas we build their technologies.

And this is the sort of really unusual aspect, I think, that we have. That we, when we work collaboratively with partners, we often play different roles for different partners, depending on their needs. And I think that level of diversity is one which we’ve been able to really grow into through the diverse set of skills we’ve brought into our team.

But it’s the bit which I’m most excited about. Because this is, I think, what differentiates us from most other groups I know that are specialists. And this is at the heart of the why don’t you just focus. We could do a lot of different things in a single topic area, but we’re doing lots of different things across a lot of different subject areas and disciplines.

[00:08:55] Santiago: And we know how to work well with specialists as well.

[00:09:00] David: Absolutely, and that’s critical. Another one of our principles and we keep coming back to these principles, this is very interesting that this is a very principled set of reasons that we have for taking this approach. But we are collaborative by nature.

So in the partners where they already had an established group who are offering research method support, we’re not trying to compete with them and take that role away from them. We play that role for other groups, but for this group we play a different role. So we want to collaborate with groups and add value in different ways.

And that collaboration element is really central. And that’s part of where we tend to find ourselves doing the work that is needed, rather than our area. We don’t want to force the area that we work in, because actually we have a diverse skill set and we’re able to often adapt and bring in other partners to collaborate where we don’t have the skills yet.

[00:10:08] Santiago: A clear example is when we were developing the responsible AI course for the Turing Institute, we brought in a group of philosophers.

[00:10:17] David: Absolutely, not just a group of philosophers, we actually had a really diverse team. We had somebody who’s at Caltech, we had somebody at Lancaster, we had a whole set of different people, and then the philosophers who came in from Bonn, we were able to bring in that team of people and actually bring together these different expertise and play that role, coordinating in that particular case.

We don’t always play that coordination role, in that case we did, and what was really empowering to me about that and why I love that piece of work so much was that it was a real collaboration and we were at the centre of it. And bringing people together with those different expertise, we’ve now taken that work on and it’s become something else for us because of that collaboration which emerged.

[00:11:11] Santiago: And a lot of our episodes on Responsible AI really build from the experience we gained from those philosophers. Learning from the experts as well. We’re not just letting the experts get on with their stuff.

[00:11:27] David: What was so interesting there, and this is where I really enjoyed that experience. The responsible AI topics, I’d been involved from the data side, from the technical side, for quite a long time. I only got interested in the topic and put something forward to Turing because there was a need from the technical side. I was working with data scientists and finding real problems in their data literacy. So I entered into that with a view that what we were going to do was going to be very narrow, and it was going to complement the sort of philosopher side.

What was so inspiring to me as coming out of that is when I started discussing these technical issues that I had identified and that I’d been worried about when I was training data scientists, I was doing work with PhDs, you know, people who were doing PhDs in data science on a doctoral training school, I was doing work with people who have real technical expertise. But I was always surprised that when I gave them tasks to actually draw insights from data, they just got fooled all the time, it was so easy to mislead them with data. I’d create these simulated data sets which just had little tricks in, and even compared to people who were less qualified than them, they tended to fall for the tricks much more.

You know, when data is unbalanced, they believe the outcomes rather than balancing it before they check the results in certain cases. And so I was really surprised at how often qualified, competent data scientists were getting fooled by the data, by simple data in different ways. That’s where I came in. But once we started interacting with the philosophers, they were exposing us to all these scandals that had happened. And what I didn’t realise is every scandal that they put to us, we said, well, okay, chances are a technical misunderstanding contributed to that. We realised that the skills that we were talking about, the problems that we were identifying from the technical side, actually were more related to the bigger questions of ethics in AI and responsible AI than we ever imagined.

So that’s where we then started getting much more involved. And now we’ve been involved in training corporates, we’ve done things with universities. It’s something where it’s a part of what we do, bringing together this responsible AI as ethical consideration, with responsible AI as a very technical, how you as a data scientist can be responsible with the data you interact with?

[00:14:25] Santiago: What I find perhaps most exciting about having worked with that group is that they are part of the team working on defining the regulations for the EU on artificial intelligence.

[00:14:41] David: Yes, and the audit process. You can’t have regulations without an audit process. And so a big part of what we were discussing with them is what might this audit process look like in these contexts when this regulation comes in, and so on.

One of the advantages of being an expert at something is it does enable you to work with other experts more easily quite often and so we had this real privilege of working with these experts in their area in this way, which was really, really inspiring and, and, and very insightful.

[00:15:14] Santiago: And just to highlight the breadth and not get narrowed into one particular example, I love the example you gave me about that little grass or weed growing in Niger.

[00:15:26] David: Well, and that’s one where my role in that was originally just to help people with their research methods. And I went in with these amazing, again, international experts, researchers who were working in specific crops in the region, we went into this village in the middle of nowhere, in Niger, and there was this weed, and I asked the expert, no, that’s just a weed, I don’t know of any particular use, it’s not my area of expertise, because of course, they studied millet, and this was not millet.

So it was fair enough that this wasn’t their area of expertise, as an academic, but they know the environment well. And what was so fascinating is then talking to the farmers on how they use it. That’s where that addition of expertise came in. And what I think is so interesting here, and I think I want to now tie together what may seem totally different, this idea of working with very basic farmer knowledge in Niger, one of the poorest countries in the world, to the work we do in the UK with the Turing Institute on responsible AI.

These things tie together. If I hadn’t done the responsible AI work, then when I interacted with partners in, actually it was Burkina Faso, not Niger, but related to this same work, and they asked me about how to take up AI responsibly. I would have had answers which were too technical, but it was the experience I had of actually working with the philosophers in Bonn which meant that actually I could sit down with them and sort of say, okay, well, you know, yes, I am now more informed to be able to actually brainstorm with you on what it means.

One of the things which I think was fantastic, which came out of that, which is so interesting, is that… it came out that one of the issues, which we know from the technical side of our work in responsible AI, that one of the problems was the biases in the data. And so we were able to identify that some of the tools that don’t work for them, almost certainly the reason they don’t work is because there is a gap in the data where their data from their context doesn’t exist. There isn’t that data bank. And so the AI training is actually all fine. The problem is that there’s no data coming out. So the question then, and this is a project that I’d love to get off the ground, is they would love to understand how can they photograph the diseases in their regions in a way which they can use them to train the apps, which could then be used by farmers to identify diseases on plants or pests. And these are such simple things, but what was so interesting to me was that it needed me to have the exposure that came to recognise, I don’t know, the, the ethics side of it, which opened my eyes to how to bring that in to other areas of our work.

And so everything’s always connected. And that’s, I think, one of the really important lessons that we’ve learned of why can’t we focus. Because actually, the problems we work on are hard, they’re complicated, and if we focus too narrowly, we won’t actually contribute the right things. It’s the breadth of our experience, the breadth of our exposure, what enables us to be able to step back and discuss with partners: what do you need now and not push to them what we’re doing. Too often that’s what you see happening. People sell what they do. We don’t. We’re able to step back and ask the question of what do you need and that’s transformative.

[00:19:22] Santiago: Can I question that briefly?

[00:19:25] David: Absolutely.

[00:19:26] Santiago: In several occasions, we didn’t push, but we encouraged groups to use our app builder for their needs. So in some ways we do push for our solutions.

[00:19:40] David: So are you thinking of early family maths now? For example?

[00:19:44] Santiago: Early Family Math, WASH-App.

[00:19:46] David: The WASH-App, well the WASH-App is different because that was part of the same project and we wanted consistency. The WASH-App was the control as part of a randomised controlled trial. And so that’s a bit different. But early family maths is an interesting one. And the question there, we didn’t push that.

[00:20:03] Santiago: Do you want to say briefly what the early family maths project is?

[00:20:08] David: Absolutely. No, thank you, you’re absolutely right. Basically, we were developing this app builder approach for our parenting work, Parenting for Lifelong Health. And we started interacting with a colleague in the U. S. who has done some fantastic work trying to get families of young children to get, instead of just early literacy, early mathematical literacy as well.

He’s got a charity in the US now called Early Family Math and we interacted with him and we discussed with him, and we told him about what we were building and the way we were building it. And we didn’t push our app builder to him but we offered it to him and said are you interested in getting your resources out, maybe in these low resource environments using our technology? It’s open source. If you want to use it, you can. We don’t have it well documented yet, but we can help you. And that’s what we did. And what was very interesting is, we just started there by just enabling him to get the resources that they had already developed available through our app technologies.

But what that’s evolving into is so exciting. And it’s certainly going beyond what we’re currently doing with the app in other ways because he’s innovating with it. This is exactly the point. I don’t think we pushed that. He sort of mentioned a whole range of different things that could work and so on and we suggested, you know, might you be interested in? It could be useful in low resource environments because… we saw what he had, getting PDFs and downloading them and looking them on the phone just doesn’t work well. But if you can present that material more nicely in an app format, people may be able to consume it.

And the books that he has, which are beautiful, fitted in perfectly to this app context. And so he has these open educational resources and putting them into our structures wasn’t something where we pushed our technologies. It was something where we asked.

We didn’t have any money to support him, but we could offer the open source solutions that we’re building for other things and say, would you like to try and use this? And again, that’s so exciting.

[00:22:26] Santiago: And in the process, he defined areas where we could improve our technologies.

[00:22:32] David: Absolutely. And we had to because there were things he wanted to do which were not yet possible in our structures. And what’s of course so fantastic about that is that’s fed back and that’s helped the projects that were using it. So his needs actually aren’t so different from the needs we had in the other projects and so everybody’s benefited. This is why I love open source technologies. If you get enough people using them in different ways and developing them, contributing to them, then they take on a whole other life of their own.

[00:23:01] Santiago: Okay, this is very exciting, but throughout this conversation, every single point you made came back to the same question in my mind. Why doesn’t IDEMS simply focus? But it sounds like we do focus.

[00:23:19] David: Well, we are strategic. That’s different to focusing. We build things together. The point is arguably we are extremely focused on a big vision, which isn’t visible to most people, which just happens to require a real complexity of different things, which all touch together.

And so from other people’s perspective, it often looks like we’re just doing random different things. Somebody who looked at us and, you know, has a business hat, a business mind and says, okay, you just have a bunch of clients and you do whatever the clients tell you.

No, that’s not what we do at all. We are part and parcel of shaping that set into a coherent whole, but that coherent whole is something which is way bigger than any component on which we’re working. They’re just part of contributing towards that same bigger whole, which relates to the open technologies we work in, which relates to the fact that actually, at the heart of everything we do is data.

The app builder we’ve discussed, well, the differentiating feature of it is how we treat elements of it as data and what we do as data. And this is where actually bringing someone in who was a mathematician who now was running a charity, he understood this very quickly, very well, and he was able to build it.

If we were trying to get other people to build it, they’d have found it hard. Because that data mindset would not have been as obvious. And so there’s elements which we have which sort of bring everything together, but in ways that aren’t always visible.

[00:25:01] Santiago: A big part of my work is creating educational resources for African universities. Even at that level, we can break down the elements into data and we’re thinking about how to process that data to create these resources more effectively.

[00:25:15] David: Exactly. And that’s something where that work is not directly data focused now, but it is about capacity building. But as we think about it, we have already mapped out how, once we have the resources to develop it further, we want to bring in our data approach, our data mindset into how this is done, to increase the efficiencies, to increase the scalability, to enable this to become something more. Because that’s the commonalities we’re seeing underlying.

We’re looking at a future, people are worried about AI. They recognize that the world of the future is a data world. But actually it’s not all big and scary about what AI can do. The fact that it’s going to be a data word affects us in so many other ways. It means, for example, that the efforts that people put in to learning how to code right now are maybe misplaced.

Because actually, even the coding in the future, especially with AI around, well, that’ll probably become a data problem, rather than a coding problem.

[00:26:25] Santiago: Well, I have seen some videos of ChatGPT4 creating software through prompts. It’s amazing what can be done these days.

[00:26:36] David: There’s a lot of good emphasis that has been put on coding. I was so privileged as a kid, I was coding my own games age 10. I’m the generation that was privileged to grow up on code. But that’s not what I want for my kids. I want them to grow up on data. I want them to be better than I was. I want them to recognize that code is like unstructured data. What you want is you want to understand the structure in the data. It’s not enough to just think of it as code. You want to think of it as data. Now maybe, maybe I should correct myself. Code is semi structured data.

[00:27:14] Santiago: Well, let’s not get into that rabbit hole. Let me perhaps finish this with a final question. You had extremely wide range of experiences while growing up and while studying and so on. How much do you think that influenced this view of IDEMS?

[00:27:41] David: I can credit the braveness to be able to take on and to do this explicitly to my time in Kenya. This is something where this is where I’m indebted. I grew up in Niger, but this isn’t about my time in Niger. This is about my time as a lecturer in Kenya, where one of the things I recognized and I realized and I learned when I was there is actually, if you have the skills, most problems don’t need true experts. They need expertise, but actually the level of skills that you need to do many tasks reasonably well is not what I would consider a world expert. So when I went in to Kenya, I was a world expert at tilting T structures on Calabi Yau varieties related to Del Pezzo surfaces.

And that was absolutely irrelevant to anything I was doing or teaching. I did manage to get one or two students to study in that area. But the fact that I was a world expert at something was irrelevant. Where I was able to add value was where I wasn’t an expert, but I was good enough. I was able to add value a lot in teaching people about data, in teaching people about linear algebra and other concepts, where actually these were able to serve them.

And from that, they were able to get skills which they needed, and they were able to then actually deliver things which were useful. And so one of the things which happened to me while I was in Kenya, I spent six years at Maseno University, and what happened while I was there is I recognised that although I wasn’t the word expert of many of these different things, I had the skills to be able to be good enough to be useful.

And that’s the key. It’s about adding value and being really useful. We are experts at certain things and I believe in certain areas over time as an organisation, we will be known as the world experts in certain areas. But what I hope is, in everything we do, we will be seen to be useful.

That’s why we shouldn’t focus. Because perfect is the enemy of the good, is a very common expression. We need to be useful. And actually there’s too many people who try to be the experts and try to be the best at something. We don’t need to be the best at many things, we just need to be good enough to be useful.

[00:30:29] Santiago: Right, that is a fantastic, very, very interesting perspective of businesses and impact. And I’m afraid we’re out of time, but we will touch on a lot of this in future episodes.

[00:30:47] David: Absolutely. And maybe just to wrap this up, because I want to come back to the original question. The original question is, why don’t you focus? And the simple answer to that, I believe, is because we want to be useful.

[00:31:05] Santiago: Great. Thank you, David.

[00:31:09] David: Thank you.