← Humanitarian Frontiers
Humanitarian Frontiers
Nowhere To Go but Up: Future Trends of AI Use in the Humanitarian Sector
00:46:52
Loading summary
Transcript53 lines
- [00:03]
Chris Hoffman
Welcome to Humanitarian Frontiers in AI, the podcast series where innovation meets impact. In each episode, we dive deep into how artificial intelligence is reshaping the future of humanitarian work. From enhancing crisis response to making aid delivery smarter and more effective, AI is opening new doors in the way we support communities in need. In this series, hosts Chris Hoffman and Nassim Motelaby bring you thought leaders from academia and the tech industry to discuss not only the vast opportunities AI offers, but also the ethical considerations and risks we all must navigate. Join them on this journey as they explore AI's potential to transform lives and address humanity's most pressing challenges.
- [00:48]
Hey, Naseem, welcome back. It's been a month. So happy to see you back. It's so nice to have you back in the office and around. How were your travels? Did you have a good break?
- [00:58]
Nassim Motelaby
It was an amazing break. I had a very good digital detox, like no laptop, no AI. Except when I saw AI Summit on the news and I was thinking fomo. So I'm happy to be back, kind of diving in and seeing what's happening in my own organization. And also there's a lot going in the world. So kind of curious to know how is that affecting us?
- [01:24]
Chris Hoffman
That was probably the most significant month in history to be digitally detoxing. Let's be fair. Like that was a crazy month that we just went through. Still feeling the effects. I think I've still got another 18 months to get rid of the effects of what the last 30 days have been. But you know what, we're joined here today for this episode. We're going to be talking about trends, what's really where we headed, what's the thinking, what are people talking about? And we've got two great guests with us today. So firstly, I want to welcome Nana Gamkralidze from the ifrc. And it's really great to have you here. Nana based in Budape, Budapest. And we've got Karen Masal here from Data Friendly Space, but also from the HTH Network, which is an amazing group of like minded organizations and partners around the world that have come together to talk about how to make the humanitarian sector more efficient, more effective and more impactful in their work. So Karen and Nana, great to have you both here.
- [02:16]
Karen Masal
Lovely to be here.
- [02:18]
Nana Gamkralidze
Thanks for having me here. Chris, Nassim, Naseem, I want to just.
- [02:22]
Karen Masal
Say that I was at the AI Action Summit and you should have fomo. It was a fascinating event.
- [02:27]
Chris Hoffman
So I mean that invite list was crazy. Like everybody who is everybody who is everybody was there, it was amazing.
- [02:36]
Karen Masal
It was a great time to demonstrate the work humanitarians are doing also through various projects.
- [02:42]
Nassim Motelaby
So, yeah, so maybe we should start there. What do you think came out of that summit and what does that mean for humanitarians?
- [02:52]
Karen Masal
I think from a more political perspective, maybe for Europe, it was an opportunity to reflect and prove that Europe is a good place for innovation and that Europe should not only be seen as A or eu, shouldn't only be seen as this regulatory machine, that there is this appetite and initiative for innovation. And hopefully for the humanitarian sector today, where we are grappling with various donors, including the US stepping back, we should have this sense that there are alternatives and Europe is taking this seriously. For humanitarian organizations, there were a few that were there presenting their work and their tools. Again, really great opportunity for the humanitarian sector to show that we're not lagging behind. And usually when we're blamed for not being innovative, that's actually not true. And it was a great moment for humanitarians.
- [03:48]
Chris Hoffman
That's really cool. I mean, just the announcement by France after, at the end, about the investment, et cetera. Because I don't know how many of you saw the meme a few weeks ago of the bottle cap in Europe. You know, how they now have the bottle cap in Europe that won't come off the top of the can. And then it had like grok and AI happening in the US and all these things. And then it had Europe with its bottle cap. And that was like the comparative innovation meme. And I was like, oh, that really hurts. You know, as you're sitting over here watching that. But then you saw what came out of the summit and that was really positive. So. So it was good. Yeah. I try to stay out of the meme culture as much as possible, but sometimes, you know, just it hits you right in the face. Nana, you know, ifrc, one of the biggest organizations, your nonprofits, federated groups around the world, you not only engage at the humanitarian level, you also engage at the governmental level. Right. So you're very much intertwined with national based systems, et cetera. And as you talk to national societies, right, because you're sitting at the IFRC level, you've got the national societies that are around you that you're supporting. What are you seeing from national societies? Maybe make this a two part question. The first part is you've got Global north national societies, which in general support the Global south, national societies in many ways. Right. And so what are you seeing from that perspective? So Global north to Global south and then the converse of that. What do you See, from the Global south, that can feed into the greater federated structure of the Red Cross around AI.
- [05:19]
Nana Gamkralidze
Thanks, Chris. That's a very good question, to be fair. I mean, first off, I think when we talk about AI, I think like in the current context, most of the time, especially including in the ifrc, IFRC network and with the national societies, when we discuss the AI, we mostly mean, I think, the large language models, because that is all the hype currently. It's very interesting that the adoption has been very uneven. I would just say that everyone currently is just playing with it a little bit, both at the ifrc, you know, the Secretariat, and also, you know, the entire IFRC networks, you know, in all over the IFRC network, the national side is all over the world. Everyone is, I think, still testing the ground. I mean, what it could mean for us as humanitarians, how could we best use it. And also, there is a lot of discussion going on about the risks, how this could affect the communities we serve, what could be the potential risks with the data integrity. However, these are still very much on the level of, you know, just general discussions, because I think still at this stage, the adoption is very limited to a very, very specialized groups. I mean, recently I was actually analyzing this survey data and one of the questions was actually examining the use of different AI tools. And I think the majority of respondents said that they do not use any of it ever. So I think it's still a long way to go. Because on one hand I think AI is all this hype, but on another hand, I don't think that we are at least. Yeah, at least in this region, in the region of Europe, we're fully aware to that the institutional capacities of organizations, NGOs, and also the ones similar to Red Cross, national Red Cross and Red Cross societies are very, very different. I mean, the. And also the IT infrastructure they have is very, very diverse and different and varied.
- [07:40]
Chris Hoffman
I can totally imagine, I mean, the scene from your perspective, obviously, coming from what is seemingly almost a digital first kind of organization, right, at wfp, and then thinking about the trends that we're looking at, right? We've talked about agentic AI in the past. We've talked about what are the levels of, you know, data that is required to actually allow these tools to do the things that they need to do. So now that we've hit this kind of brick wall of decimation right across the sector, right? Today I saw a figure from USAID which was the total amount of USAID cuts was 262. And that doesn't include the Netherlands, 300 million. That doesn't include the European Union, probably 2 billion this year. Doesn't include Finland or Sweden or all the others. Right, so Switzerland, amongst others that have cut funding. So when we're looking at this time of austerity and we're looking at technology, most people that are foundational in the sector would say, let's go back to the way that we do things. Let's go back to our core business. Let's get into Land Cruisers, deliver the food and leave. But knowing what we know today, knowing what the scale of the crisis is, what are organizations or what are you hearing as you discuss with not only internally at wfp, but the greater UN system, who's being decimated as well through these funding cuts? What are they saying about AI? What are they seeing as the path for AI in these coming 12 to 18 months?
- [09:11]
Nassim Motelaby
Thanks, Chris. And what Nana is referring to in terms of organizational capacities and readiness, I would say in terms of adoption, I think it reson even with organizations like wfp. But what I'm seeing is a culture shift. I think for a long time we've struggled in the UN sector to adopt AI analytics. I understand when we speak about AI these days, a lot of times we are talking about large language models or solutions that are ready to be used, and that's a very attractive area. But it has also brought in the conversation on what are we doing with analytics. We see analytics in sectors like health care a lot. We know the data, we know the problem, and we know what we're looking for when it comes to, let's say, anomaly detection, disease detection, prediction and all of that. But for us, for a long time, we've been reacting to crisis and we always wanted situational awareness. And with that, AI analytics seemed kind of out of place a lot of times. But coming from a preparedness perspective, bringing a lot of the work that has already been done in research institutions in terms of forecasting, predictive modeling, food insecurity forecasting for us is very huge and very important. Bringing that again into the conversation and thinking, okay, how can we build on top of this and expand our capacity so we can actually be prepared for the upcoming crisis and also know the needs and know what we're dealing with? I think just understanding the gaps within the organization itself. I spoke about efficiency earlier in our, in our conversation. I think it's very important to know where are we wasting our funds and money? I think this is partially related to those analytic components and it requires a lot of data Crunching, I think that's the readiness component that I think is very important. A lot of our data is still in notebooks, in notepads. A lot of it is even not there now we see a transformation in terms of data and data processing, data governance, and the culture shift where we're actually working fundamentally in the organization to say this is important, especially given the funding cuts, especially that we lack resources. And this is good, this is a good shift. We see that this is trickling across the organization. And actually that leads me maybe to a question around data and data readiness. Maybe Karen or Nana, maybe you have some thoughts around this and how can we actually be better at this in the humanitarian sector?
- [12:03]
Karen Masal
I can go first. I mean, I also want to start this conversation with a disclaimer that we're recording this episode at a time where the sector is struggling to, to say the least. And I almost came to this podcast thinking, is it ridiculous we're talking about AI in a, in a, in an environment where we can't even deliver the basics to humans?
- [12:25]
Chris Hoffman
Yeah.
- [12:25]
Karen Masal
And I think what I reflected, or almost convinced myself, we need to keep talking about it. And we as, as one of the data frame space, as a builder of humanitarian AI, we are under more pressure and a bigger obligation than ever before to show that artificial intelligence actually brings us 100% and that artificial intellig actually be built in an affordable way for the whole sector, not just three organizations.
- [12:52]
Nana Gamkralidze
Right.
- [12:53]
Karen Masal
Nassim. It's a cultural shift also for us as builders of AI, that it's not enough to say it's bringing efficiencies, we need to show it is. And we are doing a lot of number crunching at data friendly space about how we also do data analysis in the humanitarian sector and data preparation of. Before we developed our GANP platform, which is our generative AI initiative for humanitarian data data analysis, how much funding we spent on data tagging, data structuring, this data preparation. I'm not exaggerating and I'm not even ashamed to say it, that before we started using our generative AI based tools, for instance, it cost 30, $40,000 to put together a comprehensive humanitarian needs assessment that we would launch monthly. Today we've managed to cut it down to about three, $4,000.
- [13:52]
Nassim Motelaby
Right.
- [13:53]
Karen Masal
And it's not great to say it because. But that was the situation, right? We needed data taggers, we needed junior analysts, analysts, senior analysts, quality controllers. It was incredibly heavy in terms of resources. So if we are not as data analysts and humanitarian AI developers able to show Efficiencies. I think this is not the time for this conversation, so we need to be able to do this.
- [14:19]
Chris Hoffman
Nassim, Nana, just really quickly and Karen. So we look at Fuse now has been shut off, right. When we talk about preparedness, IPC had been putting out all kinds of new initiatives on collecting data around, around predicting famine, et cetera. And so those are gone or they're completely degraded to a point of kind of ineffectiveness. Right. And then we look at Gannett and Karen, what you're doing with Gannett and the ability to kind of assess certain situations and be able to showcase and report on what's going on, situational awareness on the ground, et cetera, and what's happening. But it seems like our predictive toolbox has been thrown out the window sort of in some way. Right. WFP has some pieces to it, right. With, well, it's not called VAM anymore, but with, you know, the volunteer, what was it? Vulnerability Assessment Mapping Team and, and so on. But still it feels like we were, we were on our way, but we got three flat tires and we only had one spare. You know, it kind of, it feels like that was where we're at right now. So I mean like at nfrc. Are you guys talking about a step change? Is anybody discussing taking a plunge? I just got off a phone call with another individual and, and he's like, look, I. There's two things I want. I want no drama and I want to do things okay? And he's like, I just want to get the things done. So there is this pining from the private sector and from nonprofit technologists that are out there that want to help. It feels like there's a stop, it's a leapfrog moment and we're missing it almost. We're almost missing it. Maybe that's a fair thing to say. So how are you feeling in the ifrc? Do you think the conversation is moving fast enough or is it still kind of, you're on a three year trajectory to kind of figure out where we're headed?
- [16:06]
Nana Gamkralidze
I think, to be honest, I think currently the situation in general, everywhere, it's very fluid. Everyone is kind of trying to figure out what's. Because I think one of the reasons is that because world is changing just so, so fast and like things change overnight. And also another thing is that because of how rapidly AI is changing, it's you. The things that we were discussing yesterday could not be like that relevant anymore tomorrow. But I mean, definitely it's one of the things that I Mean, I could not talk for the entire organization, of course, but it is the thing that many teams are currently discussing and many initiatives are kind of popping up. And the thing is that it was very interesting when you mentioned that probably a lot of people in this, the aid and humanitarian field are saying how to respond to the current situation. I mean, should we go back to doing what we do best? But I think that that option is kind of out of the window now because I mean, the cat is out of the bag. I don't know if they. Wait, this is how they say it in English. I mean, we cannot really go back to where, I mean we cannot ignore, if we ignore this situation, it will not just go away. So I think that ye. And now more than ever, I mean two things. I think the efficiency and the way we work with data are going to be like two very important things for this sector to survive. And yeah, leveraging and kind of harnessing AI by the humanitarian and development organizations, it could even determine their survival in the long term. I mean this is at the moment how I see it. And I think that on one hand, because humanitarian data, as it was already already mentioned, it's very, it's sometimes messy, it's incomplete, it's again, you know, it's scattered in many people's notebooks, flip charts, meeting minutes, minds even. So this is why I think adoption of AI does not. I mean it's. AI models struggle with these conditions often. But also AI could be the solution to this. I mean it can assist so well with structuring the data and the, the multimodal AIs beyond the text only Intellig. It could solve like this whole area of problems related to, you know, it can process handwritten field reports, scanned documents, images, video, audio at the same time. So this could really kind of address so many inefficiency errors and also address the issues that the we, the humanitarians and development professionals have been facing. And we have so much data, but there is just not enough time to process and structure and work with us.
- [19:04]
Chris Hoffman
I mean on the multimodal thing, because I think it's a really important point, right? Because we, we are 90% of our conversation sort of sits around the text based stuff, right. So as we move into trends and you know, we want to talk kind of about those trends and there's some different pieces I think that we want to touch on throughout the conversation. Open source, digital, public goods, all these different types of things. What does that mean in this space? We can get into AI agents a little bit more deeply. But this multimodal piece, I feel like there's, it's, it's an added complication that technologists might understand. But to be able to express that out to others that are making decisions, that's where we kind of start to get lost. Right. The text piece is super easy. But when you start talking about computer vision, you start talking about satellite imagery, you start talking about mixing all of these things together, you know, mixed, mixed media into to one thing. It can be very complicated and very scary because that's a lot of other things. I mean, Karen, when you talk to some of the humanitarians that are, that were in Paris, right, and some of the things that they were, they were presenting, were you seeing multimodal or were you seeing kind of more on the ground implementation kind of things? What were the trends that you were seeing that were already being done?
- [20:19]
Karen Masal
Yeah, I think there's quite a push towards the more multimodal AI approaches. I mean those that are using both satellite, textual, different risk indices in their different crisis overviews and analysis. I mean, I think there's a lot of non technologists also listening. But I think the reason why we need to move towards multimodal AI systems is think of it as you as a human, you're walking around, it doesn't even have to be a crisis situation. It can be a peaceful day on the street. How do you feel? Take decisions. You take decisions based on your surroundings, right? What you see, what you hear, what your cultural background is, what you've experienced before, what you might be reading on a traffic sign. So really as humans, we take decisions using various information sources and formats. We're not just reading. And I think the more we can replicate and try to bring these into our AI systems, the more effective and more nuanced we think as humanitarians. So for instance, on Gannett, on the situation hub that we have, the big focus is on textual data, but we also have a crisis overview that focuses on some risk indices, some key statistics, some geospatial data. So while we might not produce or analyze it all, at least we are empowering the partners that do provide us with these data, use their APIs, bring them into sort of a one stop shop up for data. So they are much more complex ways of doing it. But there are other ways of bringing information into, into one place because that's how we as humans take decisions in, in real life. So, so why not bring that into the artificial intelligence systems?
- [22:08]
Chris Hoffman
The more I see what everything that's going on with data at wfp, right. It's, it's becoming such a data driven organization in many ways, right? From supply chains all the way to, to vulnerabilities to different things. And you guys have taken a big step, a big leap as an organization to have someone like you, yourself in the position that you're in there and then your opportunities to bring in other people to support you through the system. So from an organizational standpoint, how does that translate? Kind of what Karen said, but also what Nana was saying. How does that then coalesce at an organization like wfp?
- [22:47]
Nassim Motelaby
Thanks, Chris. And I'm just listening and really taking notes. Maybe I can't speak from a WFP perspective. Some of these challenges that I want to kind of talk to, I think maybe it's relevant to all of the humanitarian organizations going back to this opportunities around, okay, multimodal AI and just how can we benefit from it? There is a grand opportunity for us, but I don't think the problem is the technology and rather our humanitarian principles and the way that we humanitarian organizations, we have set up our structures and processes. For example, when it comes to imaging and from computer vision to detecting faces to all of that, we have strict rules and policies around beneficiary data and faces and so on. So there are so many opportunities out there that we tend to actually say, okay, we can't use that or it's going to be difficult to use because we have strict policies around personal data. Right? It could be your health records or it could be your face, it doesn't matter. So to me is that, do we need to kind of have a transformational shift around how we define these policies? Is it supposed to be no, we cannot use this technology and strict red line and let's say flagging it as a high risk use case or should we actually consider high risk, high impact use cases? But in that sense, how do we protect the people we serve? How do we create better policies and regulations for the protection of their digital rights and for their personal data? And this is not a new conversation. It's been there for years and maybe not even, maybe decades, when we talk about techno colonialism, when we talk about personal data and so on. But with AI, how can we actually enhance our humanitarian innovation with some of these tools and technologies that we have and we can benefit from safely, responsibly and so on. So I think we need to kind of revisit some of these policies and principles to facilitate the use of these technologies that could actually benefit us. And on a second note, I think when it comes to the use of AI, what we're challenged with is the history or the structures in terms of collaboration and partnerships with universities, with private sector and within ourselves. So for a long time we've been using open source data like you mentioned, fews Net and so on that now are shut down. We have our own data. And I think what is valuable is the methodology we deploy to use that data to create solutions. But also what are the humanitarians or the humanitarian community doing to share this methodology, to utilize the resources we have and we have developed and how can we actually go and be more active in the research community? And I think each humanitarian state organization has a research institute and sometimes it's just considered as research and it's never considered a solution to a problem. I think we need to just revisit how we as humanitarians want to do AI and technology instead of just being mere users, you know, top down. Okay, the solution is ready. I'm using it. So on.
- [26:19]
Chris Hoffman
I love it. That really dovetails into the other thing that I wanted to bring up, which is this idea of open source versus digital public good versus, you know, common use. Right. So there's really nothing within the humanitarian sector that is a digital commons. And not always is that that terminology familiar, but this idea that there is a set of code that you can access and utilize or a set of data set, you know, data sets, we have HDX and things like that. But, but you know, these, these, these commons, is this opening up a space for that type of thing? A, I guess this is a question to anybody that wants to answer. But B, my other question to that is then associated with, is that also effective? Because once things are open source, digital commons, et cetera, that means you still have to have people inside your teams that are able to manipulate that and use it and work with it effectively. So that adds a whole other layer. And so are we in a space where social impact business can take a much stronger role because they work directly with nonprofits and are able assist them like DFS or HumanityLink or other folks? Where do you see that trend? Where do we see the sea starting to move? Is it moving towards stronger partnerships, public private partnerships? Or is it moving towards we build for everyone and then everyone can use it if they can afford the staff to utilize it?
- [27:46]
Karen Masal
I can maybe take the private partnerships. I mean when we look at all the open AI models that are out there and the debate around whether smaller companies or NGOs like Data Friendly space or like many link or whoever's building technologies that is it helping to challenge this big tech dominance? Is it helping to level the playing field? I think today organizations like Data Friendly Space are able to build truly state of art tools for the humanitarian sector without massive compute resources, without these research institutions. We are an NGO of about 2020 data analysts and engineers and we're really able to keep up with the really rapid AI advancements and testing new models as soon as they come out, assessing their suitability for the sector and humanitarian sector's use cases. So in that sense we're definitely benefiting, I think the whole sector is benefiting from it. Of course, we're not naive and there are a lot of other aspects that create inequality amongst humanitarian organizations. So it's not just enough for us to put that open digital public good out there like Gannett is. We know that there are vast differences in organizations, data literacy levels, skill sets, even the data internal data that they have available. There are a lot of local organizations that don't have data teams that collect data. So open data is what they go by. So I think all in all it's giving organizations that want to build humanitarian technology the ability to focus on what they need to be good at, which is understanding the domain, understanding the challenges of humanitarian organizations, understanding what it takes to level the playing field so that local and international organizations in the future, future might have equal access. So I think in no way are we going to be able to compete with the massive tech companies out there in terms of general technology. But our aim at least as dfs, is to compete and really be strong when it comes to providing humanitarian technology for humanitarians for a very specific, domain specific use case. So that's kind of my five two cents to this question.
- [30:09]
Chris Hoffman
I mean, Nana, do you, do you see? I mean I look at, for example, 5, 10 at the IFRC, right? Their GitHub is relatively open with the things that they've done, but you still have to be able to deploy it and host it and things like that. And then the same for you at wfp. I mean, I would suspect that most of the stuff lives in wfp, right? It doesn't. And it's really tools that are associated with WFP's implementation in and of itself. And so maybe just looking at both sides of that, you look at hugging face, you know, and you, you start to look at all the models that people are putting out there. There's a lot of opportunities for us to be able to utilize different models, different, you know, different things. There's, there's a huge push of Llama being open source, et cetera. But from a humanitarian perspective, what's the risk?
- [30:59]
Nana Gamkralidze
I think on one hand, in terms of kind of democratization of AI. Yeah, this presence of open source AI models, on one hand it's great. It allows humanitarian organizations and especially the small NGOs and local actors to kind of customize the AI models to fit their own needs. On one hand, as it was mentioned, many humanitarian organizations still lack structured data and which is essential to effectively leverage AI. However, the thing is that even if the models, AI models are open also most organizations still rely on these tech giants for their cloud infrastructure or the computing power. Access to open source does not necessarily imply independence having your independent model. So yeah, I think think and also humanitarian organizations, I mean, lack technical expertise very often. I mean many teams don't even have dedicated data scientists, let alone, you know, the engineers. That is, I think on one hand, I mean real AI solutions, I think it's, of course it's essential and they are coming in the humanitarian context, but they will I think take a few years to kind of develop because you see a lot of quick like nice pilot projects that seem very effective. But I think more thought needs to be put how they will be scaled on, you know, more kind of on like big, like scaled up rather than, you know, remain this small success stories of like one very successful and NGO reach in resources.
- [32:51]
Nassim Motelaby
That's super interesting. I think the question of decentralization of AI and democratizing quote unquote or centralization of it and the capacity is very important. It takes me back to question why do we want AI to begin with? Right. And let's say if we want it for situational awareness, if we want information on demand. Right, on demand. Information in response to a situation. That's when you need to have access to data, right. And you want to have a tool that works for you to find the information or the strategy that you're looking for. So that's one scenario and the other scenario is when you want to have analytics and insights for policy making and strategy. Right. So what is the best action for an organization moving forward? And then when we look at the humanitarian landscape, we can obviously tell that we need to speak to one another. Oftentimes we, for a long time we've talked about duplication of humanitarian efforts, organizations doing the same thing in the same region. They don't even talk to each other and oftentimes it's because they don't even have the information right next door. It could be unhcr, it could be unicef, it could be any other organization and I don't know their programs. So being able to have open source data, I think it should be considered for those type of, you know, cross information sharing or information flows from different organizations. But when it comes to capacity, Nana, what you mentioned, I think, I mean WFP does this to an extent where we enable data processing so it could be ready to be uploaded to HDX or others. And I think a lot of organizations do this, but generally our capacity to share that data or utilize it is very, very low. And I think a lot of this information is still not on demand. We talk about dashboarding and creating dashboards for information access and the beauty of LLMs is because of that on demand information retrieval with a lot of the unstructured, chaotic data that you have in your organization. If you could do that for many other data points that you have, it will be amazing. But I think it's because of the lack of knowledge of how we can use our data that, that LLMs are very attractive now because it's giving us a little bit of, you know, hope that we can actually use the data. I don't know, I don't know what you think.
- [35:18]
Chris Hoffman
Well, what happens from listening to everybody talk? It moves into that next kind of thing. That's the next trend that we want to discuss, which is agents, right? So the opportunity to overlay your data sets and link your different pieces together and allow the soup to be cooked on its own kind of thing, right? It's kind of the, I mean the simplest way to kind of put it right, because agents are, are the ability to interface with all the different components depending on how you structure it, right? All the different components which are part of your stack and to be able to start to bring together those conversations based on your question or your query, bring together the data sets and make, make some sort of suggestive answer to, to whatever you're doing. We see that with deep research, you know, perplexity and, and others where they're, they're able to really get a number of different sources and bring it into a much more concrete, succinct answer. So agents, I think we know it's a trend, I don't know about AI agents in the humanitarian sector anytime soon and being able to do that. But, but I do feel like that's, we need the help of the agents, I think, because otherwise I feel like we're going to be structuring everything into silos in and of itself anyway. And so by having something on top of it that's able to glean information, know what we don't know kind of things. I feel like that's such an important trend that we're not ready as a sector yet to really dive into. We're getting close to closing and there's so many other things that I wanted to talk about. And I know you too, Nassim, but I've been deep in this. You have been deep out of it for the last month. So I mean, I want to really leave the last question to you to kind of what's on your mind? What are you thinking about? What are you when you come and see all this going on, what trend? That you were silent for those couple weeks, right. You were within silence, you know, and now you walk back into this, you're like, oh, I never thought about that. Did some new things spring to mind since you've been away?
- [37:22]
Karen Masal
I mean, I think I'm going to really be boring about this and not come up with some outwardly AI applications. But I think what hopefully the trend will be more towards sustained uptake of AI tools. I think we are still very much in the hype cycle, stuck in the we want to test tools and use tools because they're there and then we can say we've done it. But actually when we look at most of the humanitarian AI tools today, then particularly those that are more sort of digital public goods is that have not as users and also as developers been able to maintain user uptake. And we're doing quite a bit of research at DFS around why that is. How do we keep data analysts and data natives or just humanitarians that care about data, how do we keep them engaged, not only crisis but also during peacetime, whatever that means in this context. So I think one of the biggest challenges for us and opportunities is how do we keep people from coming back and giving feedback. I think if we get there in the next couple of years where we can say we have sustained use, good use of AI driven tools, I think that would be pretty futuristic already.
- [38:43]
Chris Hoffman
Yeah, absolutely.
- [38:45]
Nassim Motelaby
On my end. I totally agree with Karen on the sustained AI use and kind of investing in the right tools. Right. I think knowing your needs, knowing the capacity that you have in your organization to find the tools that work for you, it's very important. But what's on my mind in terms of AI use cases, I guess it's not much actually technical and technology driven, but the question of business models, and I have heard this a lot, that that there is not a business case for the humanitarian sector in AI and this bothers me and it is probably the reason we cannot move forward and progress in AI. And when I say this is because a lot of those corporations, and I mean tech corporations that want to just utilize their humanitarian AI for good use cases, to have a flag up and, and show off their AI for good intentions when it comes to investment, I don't think it's there. Right. And I don't think we have a clear business model where we say, okay, my investment in AI technology for these humanitarian organizations is actually going to have a return. But that return, if it's not numbers, which is what they're looking for, then what? Right. So there is always just this limited contribution that probably leads to maybe tax cuts, I don't know. But that's just me being a little bit, I guess, a little bit pessimistic. But I wonder what does this mean for our sector?
- [40:24]
Chris Hoffman
Yeah. And Nana, I mean, Red Cross has probably one of the strongest fundraising pedigrees, so to say of the nonprofit sector around the world. Right. And your ability to fundraise both from individuals. Individuals, but also your business models through blood donations etc, that extend across the world. The ambulance services, there's a lot of different things that the Red Cross provides. Right. And when you think about that and you think about the future and not necessarily low hanging fruit, but the trend that's happening, do you see one part of the Red Cross as being much more ready to move forward? Is it the blood donation groups? Is it the fundraising side of things? Is it ambulance tracking? And, and you know, but is, is there something that, that's there that the Red Cross can really latch onto to really start to move things, move the needle in many countries because you get immediate scale. You know, if you have one central tool, you've got, you know, many different national societies that can take it forward.
- [41:31]
Nana Gamkralidze
Thanks. Thank you, Chris. That's a good question. I mean it's, there is no easy answer to that because once again, IFRC network just includes national societies from all across the world. And this national societies are just so different that I don't think that there is like one thing that everyone excels at. But I mean, I think we can, when we can look at it in more like clustered regional perspective. You know, specific regions are maybe more kind of championing in the area of blood donation services or others are more in the area of disaster response. I guess disaster response is like very generic. It's kind of the bread and butter for the Red Cross and Red Crossing movement. If you cluster the regions and Sub regions. Yes. There could be kind of the collaborative initiatives that could be kind of brought together under a single umbrella, but otherwise as a network, it would be very difficult because once again, it's very diverse.
- [42:40]
Chris Hoffman
Right.
- [42:40]
I think it's been a great conversation today. I've really enjoyed having both Karen and Nana here and having you back as well. And seems like the trend is we all kind of feel like we need to be pushing hard, but we also need to be kind of pragmatic with understanding what the pieces are. I mean, any, any final words from you, Karen, on this?
- [42:59]
Karen Masal
I really liked Nassim's point about the business model for, for AI in the humanitarian domain. And, and I know this is an incredibly probably negative note to start end on, but I'm going to be a bit controversial and say that we cannot forever say that AI use in the humanitarian sector is not going to change jobs. It's not going to put certain roles out of work. The reality is that if we want to be able to be more efficient in some areas, more efficient systems are going to have to replace more manual tasks. Like the other day I was on a panel with somebody that says AI helps to take the robot out of the human. Maybe it will help to take these manual tasks out of our payroll and put them into a much more cost efficient place. And it doesn't mean that, that people will lose jobs, but it will change the way we add value to our organizations. It's happening everywhere else. I think it would be naive to think it won't happen eventually in the humanitarian sector. So I think the only way we can start to see cost efficiencies and savings is not to slap AI systems on existing staff structures, but actually figure out where adjustments can be made. So I think that's kind of the stark reality of where I think we need to go if this was to happen.
- [44:19]
Chris Hoffman
Yeah. Nassim, I just, I want to thank you so much for, for joining on this podcast. We've got a few more to go through our 10 podcast series here at Humanitarian Frontiers of AI. This has been an amazing session. Nana, I want to thank you so much for coming on today, all the way from Budapest and Karen from Tallinn in Estonia. This is our first Estonian guest. So that was super, great, great, great to have you here and really exciting to, to learn more. I think we could have talked for another hour on so many things. I mean, there's so much here and so much to, to, to parse through, but I love the idea of the positivity. So thank you all for thinking positive. Right? We're getting there. We're getting there. Nassim, you want to close us out today?
- [45:03]
Nassim Motelaby
Thank you all for joining. It was great meeting you. Karen and Nana. Really looking forward to see what we can achieve together in the future in the humanitarian space, but also kind of exchange use cases. And speaking of positivity, what are the creative solutions that we're coming up with and that we can share across the sector? And I think that's we're talking about open source and sharing the knowledge. This is something that I've been learning in wfp and I think it's the most valuable thing that we can do you. And thanks, Chris, for having me.
- [45:36]
Chris Hoffman
No, well, you're here with me every time, so.
- [45:43]
Nassim Motelaby
It felt like I'm a guest this time.
- [45:45]
Chris Hoffman
No, no.
- [45:45]
Nassim Motelaby
Well, you know what?
- [45:46]
Chris Hoffman
You're an expert too. We can't forget that you're an expert in this field. So I gotta ask you some questions. Right? I'm. We were on one phone call was three PhDs and a Chris. And I was like, you know, guys, let's. Let's be honest. Come on. I mean, I'm just asking questions. You guys are the ones that have the answers, so. No, of course. But great to have you all here and I wish you a wonderful rest of your week. I hope you have sun wherever you are. We'll see you soon. And can't wait to talk again. Thanks, everybody.
- [46:11]
Nassim Motelaby
Thanks, everyone.
- [46:11]
Nana Gamkralidze
Thank you.
- [46:12]
Nassim Motelaby
Thank you very much.
- [46:15]
Chris Hoffman
Thank you for joining us on humanitarian frontiers in AI. We hope today's conversation gave you new insights into how AI is transforming humanitarian efforts and the steps we need to take to ensure itself done ethically and effectively. If you enjoyed this episode, be sure to subscribe and stay tuned for more discussions with leaders and innovators at the intersection of technology and humanitarian work. Together, we're exploring how AI can bring real change to communities in need. Keep pushing the frontiers of possibility.