 Tilburg University and MindLabs present Tijsig Talks, the podcast that keeps you updated on the latest developments in the field of artificial intelligence in just one hour. Welcome to this Tijsig Talk. My name is Pieters Pong, I'm a professor of computer science, and in this talk we give attention to some of the aspects of artificial intelligence, particularly aimed at Tilburg University. In previous talks we often discussed artificial intelligence and artificial intelligence has got a lot of attention in the news recently. And what we see is that artificial intelligence is seen by many people as some kind of danger. We see a lot of threats being discussed in the news with respect to, for instance, the job market, to education, to privacy aspects, to the spread of false information, even criminal activities. And in this way people are starting to think that artificial intelligence is something that is to be avoided and that we should be fearful of. However, what I wanted to give attention to in this particular talk is the positive sides of artificial intelligence and that is why I invited two people here, Marlene Balford and Walter de Banen, who I will have a talk with about the zero hunger and zero poverty labs that Tilburg University is organizing. So I would like to start with you introducing yourself for a moment. Marlene, can you start, please? Yes, thank you. My name is Marlene Balford. I'm an assistant professor in operations research and machine learning at the Zero Hunger Lab. Okay, Walter. I'm Walter de Banen. I'm an associate professor in the cognitive neuropsychology departments and I'm part of the zero poverty lab. Okay. Well, and that is a very tight introduction of yourself but probably we get to learn more about you when we start talking about these labs. So these are two labs. One is the Zero Hunger Lab, which has been in existence since, I think, 2019. Yes. Yes. And then the Zero Poverty Lab has just been started. Well, a few months ago. A few months ago, yes. Okay, so let's start with the Zero Hunger Lab because that has been in existence for a while and has been an inspiration for the Zero Poverty Lab. So Marlene, can you tell me a bit about the Zero Hunger Lab? What is it? What are its goals? What kind of activities do you have? What kind of people are involved? Which kind of organization are involved? So the whole getting a boodle. Yes. Thank you. So the Zero Hunger Lab has as a goal to help NGOs and governments to help better. And NGOs? And non-governmental organizations. Okay. So you could think of UNICEF, the World Food Program, this type of organizations. And we do so. So our goal is to reduce hunger in the world using mathematics and data science. And we do that through several projects. So of course, our baseline is research for universities. So we conduct research. And each research project is centered around a PhD student and one of those organizations or maybe a governmental organization. Maybe it's easiest to give you an example. Yes, please. Go ahead. One of our researchers works with the World Food Program. And the World Food Program, you probably know them. So they're from the United Nations. And they distribute food across the world to places where it's most necessary. For example, they're active currently in Afghanistan, Yemen and many other places. But they do more. So they also advise governments. So hunger is not just not having enough food. It's not just about not getting the calories. But it's also about malnutrition, having insufficient nutrients in your food. Because you can get enough food. But if you don't get the nutrients, development and health are just not possible. So one of the things that the World Food Program does is a project called filled a nutrient gap. They look in a country or a region at the amount of money that a family earns and the amount of money that a family needs to have a nutritious diet. And if you compare them, you see that there might actually be a gap between them. And based on that gap, they can then advise governments on interventions that they can take. For example, should they give out school meals? Or should they enforce these school meals with certain nutrients, fortification? That's what it's called. Now, before that, they, of course, have to know the amount of money that you need for a nutritious diet. So there's a bunch of calculations going on in there. And what they use is a mathematical optimization model to determine the least cost that you need to get a diet that is sufficiently nutritious. Now that mathematical optimization model is one of the things that we are working on. So one of our PhD students is working on the mathematics behind that. And one of the things that the World Food Program now also wants to take into account is the environmental footprint. Because climate change has an effect on hunger, on food security. So this is an important aspect. You can go for the cheapest diet, but if that is causing a high environmental footprint, you're carrying water to the sea, basically. So this is something they want to take into account now as well. And that makes the problem much more complex because you're now balancing not only nutrition, not only price, but also environmental aspects. So this makes it mathematically more complex. Now, this PhD student is working together with the World Food Program, together with also Kebgem and I who then do the implementation. So now you see the different partners collaborating, research, NGOs, and industry together to create a tool that can help the World Food Program. Okay. There's probably a lot more that we can say about that. When you were talking like that, I'm thinking about models that I also developed in the past for particular other problems. And one of the things that I found is that it's very easy to get out of the model that you politically want to get out of it. So how do you guard against that? Because you already say, well, there are lots of aspects that you want to take into the model, but probably if you want to make the model complete in some sense, there's even a lot more that has to go into it. So how do you take decisions on that? And how do you make sure that decisions that you take are in line with the goals of the Zero Hunger Lab? Yeah. So in this case, actually in all the cases, but specifically in this case, our models do not take the decisions. Our models give insight. So this model gives insight to the World Food Program how these different aspects can balance. For example, if you have the choice between low carbon footprint and low cost, there is of course some sort of trade-off between them. And we can give the insight in how that trade-off looks. The model will not give one answer. It will really use all the data that is available and all the mathematics that we can use to give insights to the World Food Program. Okay. And the Zero Hunger Lab has been in existence now for three and a half years or so. So evidently it's a success. Evidently what you're doing, it gives enough reason to let it continue existing. So has it been growing? Has it gotten more attention? So can you say a bit about that? So how does the immediate future look like? So the project that I just mentioned was one of the first projects. This was also one of the first PhD students we started out with. And then we were just a very small group. But in the meantime, we've grown to nine PhD students, six people who supervise these PhD students and support the lab. And also a group of master students who write their master's thesis, either in support of these PhD students' researchers or to do some explorative research. And usually we have somewhere between 10 and 20 master students walking around. So it has definitely been growing. Yeah. For people who don't know a PhD student there's somebody who is working on a thesis and has four years of research usually. And so there's some cost involved. So who is paying for that? Who is getting the money into the lab to be able to do all this work? So different money flows coming in. There is the university itself. So both Tysham and the university in general. The Tysham is the economics department. The economic faculty, yeah. And at the start a big sponsor for us was the Ministry of Foreign Affairs. So they supported us for the first couple of years. Still supporting us until this summer. But there's also private donors. And of course, as every researcher does we also apply for research grants. Yes. Okay. So this is a brief overview of the Zero Hunger Lab. We'll get deeper into this and particular projects a bit later. The Zero Hunger Lab has been an inspiration to start something else. Namely the Zero Poverty Lab. And actually the Zero Hunger Lab started at the economics faculty. And the Zero Poverty Lab started at the behavioral science faculty, right? Yes. But in close collaboration with the Zero Hunger Lab because it was a very nice inspiration given their success and also this kind of research they are doing. So we were inspired by that and we wanted to do something about poverty with the focus on the individual person. So we don't want to change systems per se. We want to start from the individual. We want to examine how poverty affects people. We're from the social and behavioral sciences. So we're interested in behavior of people and more specifically in cognitive functions like tension, memory, executive functions. So how can people plan their actions for the future? And there is already some research showing that poverty and all things related to poverty because it's really a multi-thesit problem in which impoverished nutrition also plays a role. So there you already see the match between the two elapses. So we see that poverty has an effect on the brain and on specific brain regions and networks that are involved in these executive functions making people actually make decisions for the short run. So they don't plan for the future. So their decisions are made for getting a reward or getting a reward on the short run. And that's actually the start for research. We want to get more insights into how different facets of poverty affect these networks. Also when exactly these effects occur. So during the lifetime we look at the lifespan perspective. So probably the effects are largest if people are born in poverty even when they are adults poverty might have a large effect on the brain. So we want to get more insights about that relationship because that insight can then provide us tools for interventions. So the brain is still plastic or there's some plasticity so the brain can adapt. So if we have interventions that tap into specific processes specific networks in the brain we can get better results and actually breaking the circle of people making the wrong decisions leading to more poverty and actually their children and also get into poverty. If you can break that circle by these interventions by providing insights to people who can make a difference also to governments, NGOs. That's actually the main goal of this lab. It sounds rather complex. I would actually think it would be fairly simple at least the way you started it. Namely look if I am in poverty or I am hungry and somebody tells me look here there's some food but if you spend this in the right way then you get more food in three. But yeah I am hungry now I am going to eat it now and the same with poverty. So if I can get a little bit of money right now because I need to spend it on something particular I am going to do it and then it sounds like a simple solution to give people enough now and enough for the future but that's probably maybe this is too simplistic a way to look at it. Yes so we just started. There are some studies looking into the effect of giving money to mothers for instance but we don't know what these people do with that money. Maybe they spend it on good food or just spend it on clothes so we don't have that insight yet. So that's also one of the research questions we have. Imagine indeed you give money to people how do they spend it and what would be the best way to spend it. But poverty is probably not only about money right? No it's that the way that you translate it. That's exactly the point. So it's not only related to money it's also related to be deprived of good food of good education of good social contacts. So it's much more than money and we should look into all these different facets and see which has the most effect on the behaviour and the way these facets have effect on behaviour is through the brain. That's why we focus on the effect on the brain because that's actually the mediator between these factors and the behaviour of people. Okay one more question. I want to get more into the technology but I've been reading for the last few years and maybe this is simply too optimistic but the people who say poverty is actually getting erased from the world. If you look at how poverty was let's say 10 years ago and 5 years ago and right now and then maybe the Covid actually had a negative effect but in general poverty is getting removed from the world. Is that correct? And if so, how is that happening? Well there has been a large improvement in the last 25 years up till 2015. Actually I think the best moment was 2018 so they go from more than 35% of the world population living in poverty. So that has improved and I think they went below the 9% so there was a huge improvement but then you already mentioned Covid but it's not only Covid it's also other conflicts like the Russia-Ukraine conflict but also climate crisis so the numbers have been rising since 2018. And of course there's a lot of poverty in the world but even in developed countries like the Netherlands there are a lot of people living in poverty not that these are extreme forms but still people live in poverty and also these numbers have been rising again so they are now above 1 million people in the Netherlands living in poverty so that's why we also first want to focus on the Netherlands and that's something we're going to talk about later I guess because we do have data for developed countries so we can get an insight about the underlying mechanisms and once we have that insight we can try to translate that to other countries that are less developed of course knowing already that maybe other factors might be there. I'm thinking the focus has probably looked we've seen improvements in the world that has helped a lot of people but not everybody and that's probably a group that you cannot help with I don't know exactly what the things are that have been happening that helped people probably a lot of technological development was part of that that people were able to produce food for instance more effectively which helps a group of people but not everybody because it only affects people for whom the food was the problem but getting it to zero poverty or zero hunger that means that you have to help everybody and it's then start the idea that the focus should be on the people that could not be helped with what has been improving in the world but now you have to find out why you can't help those people at least why they were not helped and how you can help them further is that a good translation I'm looking at both of you here yes I think so yes well of course you should make that remark but ok you're now also referring to the names of the labs right so the Zero Hunger Lab is called the Zero Hunger Lab because of the Sustainable Development Goals you may know them the United Nations and all the countries related to that have pledged on a number of goals 17 Sustainable Development Goals the second is Zero Hunger and Zero Hunger Lab the first is related to poverty so basically our labs are inspired by these first two Sustainable Development Goals with the ultimate goal to remove all hunger and poverty in the world which is very idealistic of course but we want to try to make baby steps first yeah but you have to be idealistic because if you say well 2% hunger is ok then that last 2% of people are of course I would say screwed and no you are also part of the humanity and we should help you as well so what I am because I said well this we talk now about goals and about the effects that you have on the world could have on the world but I am very much interested in the technology that is behind this so how do you by the way you already mentioned the idea of building models but if you want to build a model of course you need a lot of things you need data, you need certain technology to get that data into a shape that it is a model in which you can do predictions so can you tell us a little bit about that is for both of you of course so what kind of technology by the way you are using all the technology what do you use, what do you need how can it be improved and why can we do this now and so this time in history and of course for Wouter I have the same thing but of course you are still starting out but still you probably are things that you can tie into there as well but Mayday maybe you can start on that so there is actually a variety of sorts of technology and that is all depending on the question that you are answering so in the previous example that I gave I talked about mathematical optimization because it is an optimization problem but there is also a lot of technology that uses for example public data you might want to predict when hunger strikes somewhere and that is a lot of factors going in there could be a drought or it could be conflict and one of the other projects that we are doing is on predicting when hunger strikes based on public data, social media data for example in Somalia there is a lot of conflict going on and based on social media you can already see certain semantics coming up so through semantics analysis on this social media you can predict pretty well actually what one of our researchers showed you can predict pretty well when conflict starts rising hence hunger becomes an issue as well but actually there is lots of data science techniques public data is there non-public data and if you then ask why does this become possible now well that becomes possible with lots of public data becoming available with technologies becoming available but also with the realization that to solve this kind of questions you need people from different disciplines so also now that the interdisciplinarity is becoming more of common ground we have more knowledge in the same project and hence we can tackle this kind of questions and why is interdisciplinarity becoming more common ground? well it's becoming more common to collaborate between disciplines it's very classical to stay within your own discipline and work that out very carefully but now that we have all this data we have all these techniques we want to answer these questions we have to work interdisciplinarity can I make a suggestion there because now you say it like this I'm thinking so what I saw is I've been working in computer science for 30, 35 years now and computer scientists were really working on their own things they had troubles communicating with people outside computer science so they built applications and then people used them but nowadays you see that there is a by having tools being developed it makes the technologies of computer science more accessible to other sciences but these other sciences start using this and then they come with new questions and then computer science gets involved again to solve those new questions is that the reason that this is or what do you think? and another thing that and the things that you remarked is you were talking about social media and this is something that I am trying if I'm a technology optimist I love technology and I think it can help us do a lot of things but I see a big danger in social media and there are many dangers in social media of course but one big problem at least with social media is that only something like 10 or 15% of the people actually are involved with social media so you get limited information which feels like it is information about everybody but it's only about a limited group of people is that correct what I'm thinking in this, let's say in the scenarios that you are sketching and how do you deal with that? so we do not entirely know that yet so of course you are correct there is only a small portion of the people active on social media hence you miss out on the others so I think that this one particular study that I just mentioned about predicting conflict in Somalia showed that this small group of people gives you enough information to show the big trends that are going on in a certain region but yeah we should be aware that we might actually be missing out on things that are going on that are not affecting the people who are not on social media and I think if you talk about zero hunger and zero poverty the people that you're interested in are probably not very active on social media exactly, yes and that's a danger for sure I'm looking at Wouten because you are starting out on something yeah that's a very good point so going back to the question why now well if we want to answer our question the link between poverty the brain and cognition we need a lot of data and imaging data is per definition very expensive so people have done studies that are very small so maybe 20, 30 subjects in one study but that doesn't allow you to study things like poverty so in the last let's say decades there have been some initiatives to collect data on a wide scale and also make them publicly available so for instance UK Biobank to name one they have a database of about 500,000 people of which up to now about 60,000 people were scanned so that kind of numbers were yeah we could not imagine that a few years ago but then going back to the point of representativeness that might also be a problem in those databases so these databases become available but people in poverty might be underrepresented in these databases so we have to be very careful about that use techniques to actually weigh them more so that we really can examine what's going on in that population now another aspect why now is as Marlene already alluded to or actually mentioned is that people work more and more together from different disciplines the University really focuses also on that interdisciplinarity and on really bringing different departments but also different faculties together I think the Zero Poverty Lab the collaboration itself with Zero Hunger Lab is a very nice example of that because the research we are doing we have a focus on the brain so these newer imaging techniques the expertise should come from our side but the analysis techniques we are talking about now we actually need input from people with more expertise on that side so only when you bring these two disciplines together you can do this kind of research so you mentioned imaging techniques what exactly do you mean because let's say I'm pretty interested in as you said well we have lots of data but what is that data and especially when you talk about imaging it's probably not that you talk about people taking photographs in the streets you're talking about something else you're talking about newer imaging data so this is really putting people into the scanner in hospitals and then scanning them in different ways depending on what kind of scan you make you can get an insight about the structure or anatomy of the brain but also on how the brain functions so you can look into how these different structures actually connect to each other are the structural connections between these brain regions but you can also look at how do they work together so how they are functionally connected so you can imagine if you divide the brain into let's say 200 brain regions and you compute all connections both from structural perspective and from a functional perspective you get already quite complex data and for instance we know that these two things are related to each other so the functioning of the brain is actually bounded by the way they are structured but the exact link between these two is not clear because there are very much individual there's a lot of individual variability one way of examining this link is by using deep learning so we have had some studies trying to predict how your brain regions interact with each other based on how they are structurally connected that works to some extent but again you need a lot of data to really make that work enough because at the next phase is then try to predict the behavior of people from the functioning of the brain okay well you know so many things that I have so many questions on because I know a bit about deep learning and actually I had one podcast about deep learning about the technology what can you use it for and then you say well we have 60,000 people and went to the scanner but then I think that sounds like a lot but this is an incredibly complex problem and so first of all people you talk about poverty but the way people are in poverty will differ from person to person so in some way you have to represent that and then you have you mentioned 200 brain regions and then at some point I'm thinking 60,000 people is actually not that much but also it's very hard to get because it is a lot of course if you talk about brain imaging but just say well we make it 120,000 that enormous amount of work and maybe not even possible so good news is the UK Biobank plans to scan 100,000 people so that will improve but I agree the data is so complex that we need a lot of data but there are ways of clustering those structural connections for instance so we can look at specific networks and make it much easier still quite complex but instead of looking at all possible connections we can reduce that number by using specific techniques that actually cluster regions that collaborate with each other so we can use graph theory to describe different networks how efficient are they how they are clustered so we can use different metrics to describe these brains of people and then you reduce the complexity of the description of those brains so that's one way of dealing with the complexity but in the end you want to see how these different things are related with each other so how does these different factors of poverty affect the structure that then limits or bounds the functioning of the brain and then that affects cognition so we also if you want to communicate that with people you should be able to visualize that so that's also one thing we want to do we want to construct dashboard using data visualization data visualization to show what if you manipulate one of these factors you give money for instance to people so you have that financial aspect what effect does it have on the brain and on cognition but for instance if you provide them with more healthy foods what effect does it have so these are all the factors that we want to look into and see how it affects the brain how it affects cognition and also communicate that to the stakeholders using data visualization techniques the way you are describing I'm thinking probably zero hunger is much easier than zero poverty because hunger you cannot define in the in terms of nutrients as well amount and nutrients probably and then you have this reasonably sussed out and that is probably a lot easier to measure them what Malto is trying to do but I'm now probably overlooking certain complexities there as well so what kind of data are you using and what kind of data do you need so at the individual level hunger is indeed much simpler in that sense that it's one of the effects that impacts the brain but then if you look at a more macro skill in the case of hunger you're looking at systematic data which is a different type of complexity that also means that the type of data that you use when you're working on zero hunger are more about economic data logistics data climate data, conflict data all that sort of stuff information about refugees it's more systemic data rather than based on the individual I understand, yeah you probably it's not okay we make a change here and then things are solved because it is long term you make changes to affect the world in years and maybe even longer periods of time exactly yeah and the way we try to do that is by giving organizations the tools to do this and those tools are then based on data science by mathematics okay so maybe it's not more that you want to say about this because I want to talk about one particular project that I know about by chance because I know that and I know this because some people from my department were involved with that is that they actually work with imaging data but there's actually images of people where they're running a project can you say something about that because it's a different kind of data than what you have been talking about until now I think you're talking about detecting malnutrition in children yes, so this is a project that we do together with a German organization Weldhungerhilfe so what Weldhungerhilfe is developing is an app and the app can simply take a photo of a child and then the app is ideally saying this child is malnourished hence needs help or this child is okay because what currently organizations are doing is they need all the measurements scale and lots of other stuff to figure out the height, the weight of that child and then figure out whether they are malnourished we have the consultation bureaus I don't know the word in English but you bring your child to a bureau they get measured, checked every once in a while to check up on their health they don't have that in many countries so the app would be a way to make that very quick and easy to detect which children are malnourished behind that app is of course an algorithm that based on a photo of a child can make that prediction so then you go back to the deep neural networks again and the algorithm that is indeed what we're working on again a collaboration between faculties, Thaiserman, TSHD and then you go really into the artificial intelligence of image analysis now this field of predicting body shape measurements from a photo has been around a little bit mostly for the fashion industry or the gaming industry but this is all about adults not about children and we're looking at children under the age of 5 because that's where malnourishment has a huge effect hence there is little knowledge but there's also very little data there is no data on children with their body shape measurements it's also super sensitive data so the first step that we had to take in this project was collect data and that's what our PhD researcher has done over the past year she has collected data so far in one region but you've got to start somewhere and we can start training models with that okay but that's probably you need a lot of data to do this yes I don't see one piece these students gathering all that data but there's probably collaboration there locally there's collaboration going on, yes absolutely yeah okay and then you get of course the issue again that's what you hear a lot nowadays when people complain about AI is that okay it's trained but it's trained on this group of people and now you're excluding that group of people and we get wrong information about that group of people but because you try to apply it everywhere but that only works in a limited space yes and that's definitely a challenge so like I said we now have data from one location from Sulimania in Kurdistan which is one specific group of children but children's bodies all across the world are very different so of course the next step would then be to extend this to a larger definitely more diverse population I can tell you how I probably would try to solve this myself setting up some kind of pipeline where you say look if you want to have this for your region then these are the steps and here's our software and this is what you need to do and then you feed your data and you collect that into the software and then you get a model and that model and then we're going to test that model for you yes yes and that's then the next step going from research to a little bit more towards implementation and this is usually where we move to our partners again yes because at some point the university has to say okay our work is not done but this is better taken up by someone else now exactly with the different expertise yes yeah so how do you envision this because your first step is probably going to test okay now I'm going to put words in your mouth so let's let you tell yourself so what is the you're probably going to start somewhere and say okay we'll see how this goes but where do you want to go of course we're driven by the availability of data as we already mentioned there are some databases in the Netherlands that are very nice but they are still much smaller than for instance UK Biobank or some databases in the States so we want to start with the largest ones and try to find links as I explained before but then we want to see if it generalizes to people in the Netherlands because our focus is first on the Netherlands and then we want to make the next step but first try to see how that relationship is in the Netherlands so we need to try to generalize our findings to the Netherlands on those smaller scaled databases and yeah the one way of doing that is we're thinking of using for instance explainable AI to see what kind of features really drive that explanation so how are these different factors related to each other can you use the most important features in the database that is smaller scaled in the Netherlands so that's one of the approaches that we are planning to take I'm personally a little bit skeptical about the term explainable AI because it sounds so great so we build a model and the model does some good work but now we want to know why the model does that work and we can see if it generalizes which sounds nice but usually these models are so incredibly complex that as soon as you're going to say well this is the feature that is responsible for something then you're poodling something out of a model that has an enormous complexity and try to simplify it but that simplification usually then loses the power of the model that's what I'm thinking but I don't know what your experience is isn't it? I don't have experience on that yet so that was one of the avenues that we are envisioned to take but we are open to all possibilities yeah so okay well this is of course why we collaborate because in general we don't know enough anymore about how artificial intelligence actually worked in the past we did know 30 years ago I worked in AI but we knew exactly how everything worked because we programmed it in and now it is okay let the computer learn something and now we don't know anymore but it's working better than if we would program it in now if you say well and then we're going to pull things out that we would actually be able again to put in a program that we can understand then why didn't we do this immediately because that was impossible that AI learns something that is very complex and making it simple again this might not work but we simply don't know this is part of the AI research basically what is still possible here and these kind of questions that's also something that we cannot solve so we need to collaborate again and that's why we collaborate with people from THHD so in this in this lab you see that different faculties actually collaborate to make this kind of research possible okay so can we talk a bit about the future so why are these things going both in the kind of projects that you do and then specifically at Marlene because well the Zero Hunger Lab is somewhere and now it's going to continue and it's further growing so where is it going and maybe if you have any insight in that what kind of technologies do you see maybe things that have come up in the recent years that could potentially be applied for the goals of the labs so I think this is going more towards using all the public data that is out there so earlier you asked the question why is this all possible now well because people are generating more data and meaning that we can analyze more and get more insights in how these systems work based on the rain, based on micro systems so I suspect that this is going to be an even more important part of the work that we do we've only scratched the surface we're just beginning to use these tools and techniques so that's one of the things another thing is that data science and AI it has been reasonably well established now in commercial avenues in industry however in non-governmental organizations they are not that far ahead with data science as industry is simply because they don't have the means they're not a commercial organization they depend on donors who donate money specifically for certain projects so not for digitalization for example so I also think that these organizations are now becoming more digitized and are becoming are going to be using this data science more partially because researchers are paying attention to it and partially because within the industry there have been developments that they can now use yeah by the way I don't think that let's say you can say an industry is further ahead but I probably particular industries which I further had a lot of the industries are still also lacking behind and just looking okay is that something that we actually need to use the only reason I think that industries have to invest more and this is if their competitors are doing it they might have to do it as well so that is a driving force yeah of course Corp yeah so maybe this is too early to talk about because you now talked about your initial but where do you see this going I think the use of machine learning and AI is now mainly important in this phase because we need to get insights into all these relationships I talked about I think the next couple of years we will be still in that phase but the next step would then be to try to intervene to set up interventions and to test those interventions but yeah as mentioned we're still in the beginning phase so we start out with trying to find the relationships even looking at prenatal data because also even when the child is not born yet maternal stress for instance can affect already the baby and there's also some data available there so we look into all different age ranges see where the effect is largest or where we can intervene best then to take the next step come up with recommendations try to set up interventions that tap into the facts that we focus more that we found most important but I think and certainly at this initial stage the techniques we were talking about are crucial a little I would like to turn this back a little bit just to get something a bit more clear I'm thinking if you talk about poverty in general you can see the effects that poverty has you can define these kind of effects we want to avoid you have already a lot of data on people that are affected and on other people who are not affected and you can look at the differences you bring the brain in there so and I would think that is the brain really needed or is the brain for you really the explanation for the effects do you think it is actually needed to identify and provide solutions for poverty I think it's crucial of course okay no that's why I'm asking as said in the beginning the effects of poverty and all the related factors on behavior are mediated by the brain it is the brain that actually causes people to act in particular ways to make specific decisions if you can change specific connections in the brain for instance related to specific cognitive functions so that they make other decisions you can change okay can I then maybe do this too simple but you say well we have a group of people and they act in the same circumstances and you don't see any differences but at some point some people are going to suffer from poverty and others are not and that is because of their behavioral choices is that how I should see it yes and that's also something we want to look into is can you actually make prediction models based on all the data we have also the combination of social demographic information the brain cognitive functioning can we actually make predictions to say which people will end up in poverty will get out of poverty or will suffer in the long term of the effects of poverty they have now okay and then I just am reminded now again to the big I think the big difference between what the two of you are or the two labs I focus on is the individual in the zero poverty lab and the huge conglomerates the countries villages in the zero hunger lab that is the and both are incredibly complex individuals are incredibly complex and the macro structures in the world are incredibly complex but there is of course also a connection because poverty leads to hunger and hunger again affects the brain so I don't think we can do without each other okay yeah so you already mentioned why because this is one of the things that I was thinking about is why would we do this at Tilburg University and one reason at Tilburg University is that we represent even though we are relatively small university we represent both the technological sides of artificial intelligence techniques for instance but also the economical sides the ethical sides, legal sides the behavioral sides of science so is there something more so are there other places so I would have a feeling that Tilburg University is ideal for this but I can imagine that other universities or it's huge would claim that as well what do you think in the Netherlands for instance or worldwide we collaborate a lot for example with Wageningen because Wageningen has the agricultural knowledge and for food that is super relevant and super interesting yeah okay so there's also collaboration with other universities I would imagine that we can do a lot inside the university definitely anything more that you want to remark on that so what I think I would like to talk at the end here is a bit about your own research so what is your personal interest what are you doing in these labs or outside that because you probably have also research that does not involve directly with the lab and well can you say something about that yeah so my main focus is always to use mathematics and data science in such a way that it can help our society so there are many applications of data science and AI many of them aimed at commercial sites and that is something that has led to a lot of development but personally I'm more interested into topics that are directly related to sustainability hunger is a very good example of that health is a very good example of that poverty is a good example climate change all these sorts of topics interest me a lot and for me that has brought me to the zero hunger lab basically my main interest is of course the brain and the relation of the brain the structure of it the functioning of that brain with behavior both in healthy populations and in clinical populations of what if a patient has a particular lesion at a particular side of the brain how does that affect the functioning of that brain and then get the functioning of that patient so there are internal factors that can affect brain structure and function but there are also external factors for instance fatigue stress poverty and all related factors so it's really about the effect of internal and external factors on the structure and functioning of the brain and how that then translates into the functioning of people in their attention memory executive functioning all these cognitive functions in general okay very interesting so I think that maybe we should return to these kind of topics in a year or two and then let's see how things have progressed until then we also have within the ties you talk smaller presentations that we can probably do at some point but of course there's still a lot of work to be done is anything that you want to say to conclude this session I think we discussed a little bit also you started out this talk with the negative effects of AI right one of the effects of that AI can have is that people stay behind so generally this is a development this is something that we can use this is something that can help us in many ways but usually when you see technological developments you see that quite a big group of people has an advantage of that but an even bigger group of people stays behind cannot use it and that makes the gap between these groups even bigger and what I hope is that we can be aware of that and we can be aware of that with the result that we designed the techniques to include everybody you already earlier mentioned that only a small portion of the population is using social media for example in that sense we should be aware of it but also in the goals that we have when developing our AI tools what goal do we develop AI and we can also develop them in such a way that also the most vulnerable people can benefit from that or not also but in particular the most vulnerable people can benefit from developments in AI yes I think this is a very important remark that you make here because indeed I notice that there's a lot of enthusiasm about AI but it's always coming from the same groups of people and certain people are lagging a bit behind but they can catch up but there's probably large numbers of people that cannot make that connection and then fall further and further behind and once I heard somebody say okay so in western countries we went from the regular phone to the mobile phone then in other countries they can skip the old fashioned phone and go to the mobile phone directly that's not how it works that is what you would like to see happening but that's simply not how it works so we should really be aware of these things thank you Lauter anything more that you want to add to this no I think it is an important warning Malin really tapped into something very crucial we should be aware of so I agree completely with that thank you very much and I will probably see the two of you again in the near future thank you