← Humanitarian Frontiers
Loading summary
Transcript44 lines
- [00:03]
A
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 a 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 Motelabi brings 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]
B
Hey, Naseem, welcome back. It feels like this has been rapid fire podcasting we've been doing in the past few weeks. It's so nice to see you again this week.
- [00:56]
C
Thanks, Chris. Yes, we have been covering quite diverse topics and today we have a very special one.
- [01:03]
B
Absolutely. I think it is very special not only because our benefactor, the Humanitarian Innovation Program at Innovation Norway is here joining us today, but with some other great colleagues and so I really want to welcome them. Here we've got Teresa Marie Upstrom Pankhatov from the Humanitarian Innovation Program at Innovation Norway, where she's the head of the hip hop. We've got Sean White, who's with the United Kingdom's Humanitarian Innovation Hub, hosted by Elra, and she's the director there. And we've got Zaina Alsaman, who's with Creating Hope and Conflict, which is part of kind of the Grand Challenges Canada portfolio, and she leads that as the senior program manager. So welcome everyone. It's great to have you all.
- [01:42]
D
Thank you.
- [01:43]
E
It's good to be here.
- [01:44]
B
Awesome. Well, truth be known, we have been on panels before, but the last time I was getting asked the questions by others. So it's really nice to sit here and be able to ask you guys questions and be on the other side of the panel, so to say. And Naseem and I were talking before the call and as we were discussing in our green room, so to say, in teams, I used what we call the soft term, which is the whole sector in and of itself is in flux and that is facilitated quite dramatically because of the changing funding landscape and is because the three of you sit kind of on the donor side of the house. We thought maybe it would be good to start the conversation with what you're feeling and you can bring out your own personal feelings, but kind of what you're feeling as a donor that you're seeing what's shaping your current decision making, not just in 2025. But as we look to 2026, what are the things that are stirring currently given this space? And maybe we start with Sean on that and Sean from the uk, because the UK just made their announcement just a few weeks ago. And really, where does that sit? And how is that going to change potentially the landscape of AI and innovation in the humanitarian sector?
- [02:51]
D
Thank you, Chris, and really excited to be joining today. I mean, let's face it, it's a really grim time and it's pretty hard at the moment to see pinnacles of kind of hope amid these pretty drastic funding cuts across bilateral donors. At the same time, I guess I've been trying to tune in to a lot of the conversations that have been happening online or in different sort of networking spaces around, well, how do we think differently about the future? Is this a call for massive reform? Is this going to lead to that classic kind of conundrum of are we just fixing and dabbling and tweaking the branches, or are we really dealing with some of the root causes of this fragility that we face at the moment? And I think we could take some of the themes that I'm seeing in those conversations and ask ourselves, what does this mean for work around humanitarian AI? So a lot of those conversations pick up on themes of efficiency. And so what does that mean when we come to thinking about AI? And I think partially there, there is a working assumpt that emerging technologies, and AI in particular, will make the humanitarian system somehow more efficient with the resources that it has. And clearly they do have that potential. But I think there are perhaps some major gaps in our current understanding in terms of whether that will really hold true or bear fruit in real life. And if we're talking about saving costs, if we're talking about saving time, do we really have the evidence at the moment that AI development is going to be cost saving? Do we really have the evidence to understand, okay, if we are saving staff time on certain tasks, what can we then reprioritize that time for? So some of our most elementary assumptions around the role of these types of technologies need to be interrogated a little bit more if indeed technologies like this are going to come to our aid in this difficult climate. The other thing that I'm seeing in those virtual and in person discussions is that we need to get better at collaborating differently and reducing competition and duplication in the sector. And I feel like this is one that really resonates with, I suppose, my own personal career trajectory. And what I've seen as a major challenge for the sector, we're still not amazing at coordination or despite the best will from different actors to try and coordinate, we are ultimately sadly set up to compete four funds with each other. And now, I think is really a time to challenge that and to try and put that in the past a little bit. And AI is a great example of where there is a real opportunity to behave differently, essentially. And in some of the work that the UKHH has done, we've tried to start mapping out and we have a directory on our website of AI enabled humanitarian projects. There's a list of more than 100 projects that have some kind of public record of how they are AI systems that are being used in humanitarian response. And what you see quite quickly is that there is a lot of duplication already within that list of projects. And I would really encourage more organizations to add to that list, to check that list before they start investing in work. Because now is the time, I think that we need to really end some of that duplication. And then my last point, I suppose, is that now is also going to be a time where anything that is not life saving assistance is going to be challenged. And so we need to make the case for why we should be investing in research and innovation, including AI related work. And part of making that case requires us to also consider the costs of assurance and mitigating risks and that kind of global architecture that we might need to build to make sure that we can leverage the potential of AI whilst mitigating the risks that it poses.
- [06:55]
B
Yeah, I mean, Teresa, to you on that last point, because I think obviously we've worked together a lot on a lot of different projects and what I've seen is we're not talking about just life saving, right? We're talking about parts of the process that can be innovated on as well. And that's always been something good, whether it's procurement or whether it's different things that you're doing, greening the operations. And so from that perspective, how is Innovation Norway seeing this year? And how is. Because I'm sure you've already started to think about 26. What is that looking like for you guys?
- [07:29]
E
Yeah, one thing is what it looks like for us, another thing is what it looks like for our projects. So Norway will announce its continued support to the humanitarian innovation program tomorrow. So I'm happy for that. I'm happy that they continue to be sort of faithful donor and see the importance in innovation. What I'm wondering is how this will affect the humanitarian organization's capacity to innovate versus the need for innovation moving forwards. And I think you're right. You mentioned procurement and we've done a lot of work on how to support humanitarian organizations in becoming more innovation friendly. And one of the tools that we see is important here is partnerships. So how can we more easily create partnerships across sectors? And it seems to me as we're now sort of strained or pushed to make difficult decisions, that partnerships, that people are more open to consider partnerships and see the importance of forming these partnerships in an efficient and an impact based way way. So I'm hoping that this is a way forwards for us that we can work more on this. Another thing that we've worked on a lot is sometimes we call it business models, sometimes we call it systems innovation. But we see that any innovation that we introduce to the humanitarian sector, if it has the potential for a large impact, it often pushes on the system and requires the system to change or adapt to be able to reach the impact that it can reach. And I think AI does that. It pushes on our system. We can't just take one tool out and put AI in and things will move forward. But better it requires us to consider our other processes, our other tools. It requires a lot of capacity building or other supporting structures. So I think moving forward, looking at the systems innovation, looking at the ecosystem that the humanitarian response system is, I think will be increasingly important moving forwards.
- [09:23]
C
Thanks, Arisa. And to your point about capacity building and systems innovation, I think it's very much tied to what Sean was discussing around efficiency gains, impact evidence collection, but also what does it take to actually receive the promise of AI in our sector? And I think this goes to the question of scaling perhaps. And my question maybe Zaina, to you is what do you think are the major challenges for scaling AI or technology solutions like AI to actually see some efficiency gains and be able to measure their impact and how they can actually be beneficial to the humanitarian sector. Because I think when it comes to efficiency and collecting the evidence, it's tied to a level of scalability to actually be able to implement AI to a scale that it actually brings in the gains that you're looking for. So maybe you have some thoughts given your work with the Grand Challenges Canada.
- [10:27]
F
Sure. Thanks, Naseem. So, I mean, I think artificial intelligence is increasingly influencing various sectors and as we've noted, it's penetrating the humanitarian sector quite quickly. And so there's great potential. We've seen lots of great successes. We've also seen lots of lessons learned in our portfolio of investments across the innovation space. Whether it's to optimize clinical care or prediction forecasting tools to allocate resources efficiently. So I mean there's lots of interesting use cases that are, that we're seeing in the sector. I think for us at Grand Challenges Canada, I mean scaling is always an important topic that we like to think about. And I think one of the challenges that we're up against is ensuring that there's demand. I mean, I think there's a tendency of AI being this like new, shiny, sexy thing that's in the sector, a kind of catch all to solve all the world's problems. And I think when it comes to scalability, we need to think about end user demand at creating hope and conflict. The portfolio that I work on at Grand Challenges Canada, we're trying to serve the most vulnerable communities affected by crisis and affected by conflict. And I think we have to think about the end user demand in terms of our communities. Is this solution solving the problem that is being prioritized at the community level? And I think that's always the question. We have to go back to as great as there is potential for AI and perhaps a scalable solution, we have to think about the end users at the end of the day and is this solution solving the challenges that they are facing at the community level? I think we also have to look at the potential for adoption across humanitarian organizations. I mean, we fund a range of humanitarian players in the space from local organizations to larger agencies and organizations and international NGOs and we have to look at is there openness, is there demand from that aspect for the solution? And I think one of the biggest challenges that I've seen across our portfolio or even across the sector is we don't see enough of local talent spearheading these solutions. There's surely plenty of it out there. And I think one of the things that we learned, we recently did a paper, a learning paper on our AI investments across the last few years. And I think one of the things that was upright, upside and center in this paper was that most AI projects that have come our way are actually led by organizations in the us, in Canada and Europe, in the global North. And I think what we need to see moving forward to ensure that these solutions are scalable is more investment in local institutions that are spearheading AI investments. I actually think that's one of the biggest challenges that we're facing to ensure that the solution is actually fit for purpose and fit for the needs of the community. So I think moving forward we'd love to see more of that local investment in AI expertise at the community level.
- [13:15]
C
Thanks, Aina. I just love that you mentioned the local investments on AI and speaking to the innovation hubs, let's say, in the Global north, quote, unquote. But also you mentioned demand driven AI investments. Right. And I think this begs the question around risk mitigation or risk appetite. Maybe, Sean, you can speak to this working with the UK innovation and as the Director, how much of our investments are informed by this risk appetite, especially considering the data collection and where the data usually comes from in the humanitarian sector, but then where the AI investment is actually happening. So there's usually a disconnect, especially when it comes to innovation and the topic of digital innovation. There's always this question of agency and risk management, or not even risk management, just considering the needs, considering the affected populations and bringing them around the same table and having them as part of the conversation. And also even in that, that statement in itself is problematic because they should be naturally part of the conversation. Right, so, yeah, over to you, Sean, what are your thoughts around this?
- [14:26]
D
Yeah, it's an interesting one and I think what I've kind of observed, when we first started doing some work around AI, the conversation seemed to be essentially between those who are like, yes, we really need to lean into AI and others who are like, let's just try and shut the gate, like, this is all a bit much. And I think we have moved on even in the last year to a slightly different debate, which is that there seems to be a tension at the moment between responsible AI and innovative AI. And this is something I think we're even seeing play out in government discourse. Like even over the last year or two years we had in the UK Rishi Sunak, bringing people together for the AI Ethics Summit, trying to position the UK as quite a leader in that particular side of the AI debate. More recently we've seen Keir Starmer really lean into investment in sort of the infrastructure needed to support AI and just advocating much more that we need to be innovation leaders as well. And this to me is something that I again see playing out in the humanitarian sector. That there's, I think, according to me, there's a false dichotomy between these two things. You don't have to be either innovative or do responsible AI, you can do both. And I think as potential funders in this space, there is a role for us to champion that. It is not either or, and then to help support partners to navigate that. I think what we've seen already with our AI work is that there's a tendency for many organizations to feel a degree of fomo, essentially a worry that it's moving all very quickly. They want to be involved, they don't want to be left behind. And then that, you know, we all know that that's likely to result in perhaps not the best decision making. Right. And so through some of Elra's work to support a cohort of 10 grantees through a learning journey over 20 something weeks, the emphasis there was really starting with a problem and figuring out if AI should be an appropriate solution to that problem, and then scoping out briefs, basically not running straight into development, but helping organizations to take stock, to learn a little bit, to slow down, to build organizational readiness and to think carefully about forming the right kind of partnerships and to encourage them to be more aware of the points along a AI innovation journey where they might come into some of those stumbling blocks that you spoke to at the beginning, Chris, you know, what are the points where ethical conduct and humanitarian principles need to be informing at each stage of that journey? And equally, as you said, Nassim, there is a tendency, as we've seen with innovation writ large, to not necessarily always bring populations and frontline humanitarians into the process from the outset. I think there are now quite a few pieces of small scale research which actually show that for some crisis affected populations, their tolerance of risk might be higher than that humanitarians themselves. And I think that probably is born out of the fact that many populations recognize that the current way that decision making happens within the aid system is imperfect and influenced by all sorts of human biases and political trends. In some cases, populations or projects that have explicitly sought the input from affected populations have preferenced AI systems, or at least AI systems that are used jointly with human decision making because they feel like it might bring greater neutrality in some of our own work around participatory AI that we've been doing with our leading partners nesta, they've been developing a structured process for involving community views. And I think there's a worry from many actors that involving community seems like a massive complex process. Like how do you even bridge the linguistic divides between technologists and frontline populations? But they've been able to show that actually, even if you have a fairly low literacy, if things are explained well, then there is a easy enough way to engage people. And the value of that was really felt by populations, sort of 90 something percent of participants really feeling like they want to be engaged in these types of conversations. So I think those barriers are there, but they are possible to navigate in an unburdensome way.
- [18:54]
B
Zaina, go ahead.
- [18:56]
F
Yeah, I think I just wanted to underscore the point that Sean's making and this idea of responsible AI. I think as humanitarians, first and foremost, we need to ensure that we are being responsible data custodians. And that extends to not only principles of do no harm, but thinking about principles of do no digital harm. And these principles should not be an afterthought. And I think in the flurry of trying to develop really innovative and really bold solutions, I think we need to constantly remind ourselves and come back to this point of view that needs to be the sort of do no harm and a risk mitigation needs to be at the forefront of our decision making. And the stakes, I think in a lot of the settings that we invest in, in particular at creating hope and conflict, are incredibly high. We have risks of data breaches, hacks, exposures can potentially cost lives in areas of active conflict or fragile settings. Personal data can be exploited for targeting or discrimination. And so for us at the funding stage, when we want to consider, I mean, really any type of technology, any type of solution, but in particular now am this sort of frenzy of AI informed and fueled ideas that are out there, we really want to see the demonstrated evidence that your organization has really thought through this aspect. And like I said, it can no longer be an afterthought and it shouldn't be an afterthought. And I think that we've learned through our several years of investing in this space that deeper due diligence in this area is really critical. And consultation with expertise in the digital sort of security realm is really important. And we really want to to see evidence from potential grantees and innovators in this space to see how are you ensuring that you're protecting the end users from undue harm. And I think that is paramount. As Sean mentioned, starting with the problem, what is the problem that you're trying to solve? Is AI actually the solution to solving that problem? And if it is, then great, carry on. But I think ensuring that your solution fits your problem and that the end user safety and security is actually at the forefront of your decision making process and ensuring that innovators are adhering to stringent data protection and best practices related to no digital harm, for example, and ensuring that they're sort of taking the steps or demonstrating the steps and the know how to eliminate risk of bias, for example, you know, challenges of consent of data usage and those sort of challenges that emerge in this space and So I think I just want to sort of underscore the sort of importance of, you know, our role in the humanitarian sector as data custodians for some of the communities affected by crises, and just really underscoring that role that we play here to ensure that we're not causing undue harm.
- [21:34]
B
Yeah, absolutely. I mean, Teresa, such a rich conversation. And there's a piece to it that I wanted to raise. Again, seeing and working with you for many years. Right. It's really been almost five years. We've been working together on different projects. And what I've always really treasured about listening to you and seeing the way that you work and your team works, is that there is an idea behind pushing the envelope a little bit. Right. You do see that there is high potential in certain technologies potentially or in certain actions. There are, whether it be blockchain and whether it be with identity, whether it be with communications, and now whether it be with AI. And your latest call at the hip. And so two kind of questions that I wanted to pose to you. One is it goes back. We started to touch on risk appetite and we started to kind of build the groundwork up on how we increase our tolerance, so to say, of risk. But then how do we also allow for innovation, not just AI, but innovation in and of itself, to also happen? Right, because innovation tends to be in things that tend to be higher risk in many ways. Not always, but in some cases. So the first question is, how are you balancing that? And how have you balanced that? And what's your mindset around that balance of that piece of pushing the envelope? The second thing that I wanted to ask is you have a very unique approach as the HIP in terms of the way that you create the partnerships where you do kind of a 50, 50, where you will give 50% and then technology partners, those experts in a field, if it is a technology innovation or another innovation, contribute as well. And understanding that cost component, because we are talking about the implementation of AI in the humanitarian sector and the pitfalls we need to be careful for. But we also need to recognize that there is a very inherent cost to that. And that upfront cost tends to be much higher than the traditional ways of working.
- [23:21]
E
Working.
- [23:22]
B
Right. And so you've started to find a balance with that in the way that you do things. So I'd love to hear maybe not the genesis of where you got there, but how you're seeing that be impactful in terms of the potential for scale around your technology kind of initiatives that you fund as innovation in Norway.
- [23:38]
E
Thank you. Great questions. So the sort of balance of when to innovate and balance the risks in innovation, I think there are two key questions, and one is that it needs to be needs based on. So we need to identify key humanitarian need that we're currently not able to meet. And as long as humanitarian needs are not met and people affected by crisis are not being helped to a full extent, then innovation is not really an option. It's something that we need to engage in. And so if we've identified a core humanitarian need that's affecting either a lot of people or affecting those that it's affecting significantly, then we need to innovate. And innovate is a methodology to see if we can find new solutions and better ways of working to meet needs. So I think that's a core sort of starting point for an innovation process. And the other one is that the people themselves, so people affected by crisis need to be a part of that innovation process from the beginning to the end. When we ask people affected by crisis that have been part of our innovation projects, what are your recommendations to us moving forwards? They always say invest in local leadership and invest in training. And we see this in all our innovation projects that go well, that they do have local leadership and they do have training. And they often also include a component around financial strengthening, even if that's not the core of the project. Some kind of improved methodology for VSLAs, for example, often tend to pop up in the innovation projects that we support that are the strongest. And if these things are in place, I think our risk tolerance can be quite high. And I keep sort of coming back to what Sean and Zaina has also spoken about around data, because data is a core part of AI. And I think the lack of data or wanting more data has been this core challenge that we've spoken about since I started working in the humanitarian sector for two decades ago. Right. And now I think we can see a potential for the impact of having data and of using data in a whole different way. And a question that I keep coming back to is how can we have informed consent with people that are affected by crisis and how their data is collected, what kind of data is collected and how it's being used so that we can thoroughly say that the people that we're working with, they know what kind of data we have and they know what it's going to used for, and they know the risk associated with it, and they make an informed decision that I'm willing to make this investment, I'm willing to have my data used for access to a service or a tool or support in one way or another. These are decisions that we make daily. Right. And so I think we need to make that an option for everybody. Also the majority of people affected by crisis. And so that's the question I keep coming back to, how can we facilitate this process in a sound way? And another thing that I sort of keep thinking about when it comes to local leadership around this is the digital divide and how this is affecting the development of AI now. And I think bridging this digital divide is becoming urgent so that we can ensure that the risks that we're talking about now around biased algorithms, so on are not too entrenched in the tools that are being developed. That we get this right from the beginning because that's going to make the impact so much better. When it comes to funding, this is a challenging one. So we invest in innovative financing in two ways. We both challenge the humanitarian organizations and say that if you have an innovative financing project that you want to develop or explore, then we can support you doing that. But then we also have an element of innovative financing in our own funding scheme and that we say that if you want to scale an innovation, we require you to match our financing with financing from the private sector. And currently our definition of what type of funding this could be is very wide because we want to explore together with our partners how can we use public funding to leverage other types of funding. So this can be an investor coming in and investing in the solution. It can be in kind support from the private sector, partner in the innovation partnership, or it can be impact bonds or loans or other types of financing. It's also an element of quality insurance in that and that we're not the only one that would like to invest in this. Others see the impact and see the potential in the solution. And the fact that we've introduced this element also means that we can support an innovation project for longer. And we know that if we let go of an innovation too early, chances that we're going to see it implemented and adopted in the sector is very low. So it allows us to be that stable partner for an innovation project that is showing impact.
- [28:21]
B
I love it. I love it. Thank you so much. Naseem, over to you.
- [28:25]
C
Thanks, Teresa. And thanks, Chris. I'll take the topic on data ownership and risks around data privacy and protection. I think this is an interesting one because talked about the risk tolerance or high risk tolerance of affected populations when it comes to technology use and I think there are connections between that and data ownership as well. When it comes to the context of refugees, for example, which I was familiar with during my PhD research, there were a lot of studies where there were qualitative data collected from affected populations, refugees, and they were asked, are they okay with their data collection and being used for a certain technology innovation and so on. And you see that there is this wave of awareness around how their data is fueling organizational innovation for humanitarian organizations across the world, and they want to be part of that. They were actually asking for compensation for their data. Right. So this is something that Teresa, you were mentioning. How can we resolve this challenge around data? And I think my question to Sean would be, what progress have we made in this space around data ownership, data risk assessment, and also AI innovation? Is there any progress recently around this conversation and what do we expect when it comes to AI?
- [29:55]
D
Cool, that's a tough question and I'm not sure I am the best place to answer that. So please, there is. And Zainab, please jump in. But I think it's worth just being conscious of the fact that there is an AI data dilemma at a global level, cross sectoral level, in terms of the amount of data that we have to be able to train AI models, particularly generative AI models that we're seeing advance so quickly at the moment. And then, as Teresa pointed out, the reality of the context where we work, where we know that almost by their very nature, crises will destabilize or make data irrelevant quite quickly. And so if we are trying to design systems that can inform decision making or that can act in those contexts, that's an immediate and obvious challenge which I don't see us fully being able to account for in the way that we are approaching AI at the moment. So I think that remains a massive challenge. I do think there is a lot of discussion and a lot of positive progress in terms of thinking about developing through lines between data sets. The humanitarian sector for a long time has gathered a lot of data, and we haven't always effectively known why we are doing that or how we should be using it. So I think there is a continued push on us to ask ourselves, what do we need data on? How can we strengthen that data architecture? To what extent do we make some of that data a public good, and to what extent must we be very careful to protect data of very vulnerable populations and in particular protect that from nefarious actors who might be using that to target populations, health centers, et cetera. So there's some massive questions in the data architecture, which I don't think we are fully grappling with or grappling with at a coordinated cross organizational level at the moment. We're seeing with funding cuts, major data sets, things like the DHS survey data. We're recognizing how owned that data was, for example, by the US and how problematic that will be going forward. So so in my mind there's a lot of unresolved points there.
- [32:10]
B
Zain, I wanted to ask you to jump in on this one because I think it's a big question. First of all, Naseem, a point that you brought up. I really don't want to go down the rabbit hole, but it's such a beautiful conversation, I think, around compensation and what that compensation balance is. Because it is a give and a take already, right. Because it is give us data, you get assistance kind of relationship in some way. And flipping that around, what are the implications of that? I think that's a great conversation, probably not for today, but a great conversation nevertheless. Maybe you and I need to have have that separately. So we talked about this in the last episode that'll be airing very soon, which was a lot around the EU and the EU regulations that have come into force. Now, obviously I don't think that at least what Canada doesn't and the UK doesn't. I'm not sure if Norway has to follow certain EU regulations or not around the AI, but as we move towards this. Right. So the reality is a like this first thing that started getting implemented in February is that organizations have to train all of their staff on AI if they're going to use AI. So there's a whole other piece of this puzzle which is regulatory. And so Zaina, to you on the Canada side, because you're not beholden to GDPR or the new regulations, but some of the people that you might be contributing funds to that could implement AI are going to have to be. The reason why I ask it is not because of wanting to put you on the spot by any stretch of the imagination, but it is also us trying to identify what are the kind of the gaps that we need to start addressing in our development of programs and our development of funding calls to make sure that while it's responsible, it's also governed correctly. It also is following the regulatory kind of structures that are in place.
- [33:47]
E
Sean yeah, maybe I could add to.
- [33:50]
D
That, because I think this has been bubbling up for us as a big emerging challenge for the sector. And in fact, some of your listeners may know that we have a humanitarian AI newsletter which comes out monthly and this upcoming episode, if People sign up. I can maybe share a link with you, Chris on the.
- [34:07]
B
We'll put it on the screen.
- [34:08]
D
But the upcoming episode is about governance and so the, the way that we're starting to think about that is there's a degree of governance decision making, sort of readiness that needs to happen at an organizational level. There are governance mechanisms that need to exist for the humanitarian sector. And recognizing that we are different to other sectors, we are governed by humanitarian principles. For example, how do we bring out all of that in our sector specific governance? And then there is all of the ways that we need to. And I think we need to do this a lot better than we are currently. Connect with global legislation, global positive trends in terms of AI governance and assurance mechanisms. And we're still at the early stages of all three of those, I think. And there's a real mix across the spectrum about what is needed internally in an organization. Some organizations have really sort of fast tracked and thought deeply and started to expand their teams working on this. Others are really at the early stage of just asking how do we go about this? I think in terms of that middle layer about governance mechanisms for the humanitarian sector, the one thing that is clear to me is that it's not one mechanism that we need to support governance. It's actually a range of measures that need to be in some way not funded and supported just by the organisations themselves. So for example, humanitarian organizations, no matter which level that they operate at, I do not think they have the funding nor the technical expertise to effectively do assurance or things like red teaming to the extent that is necessary for responsible AI. That means that we have to have a slightly different approach to what we would have for perhaps innovation writ large for supporting that at a sectoral level and one that is guided by humanitarian principles. So we've just done this review of peer reviewed and grey literature in partnership with University College London. And you know, one of the things that kind of came out of that is that amid the peer reviewed literature, firstly it's to the point that we were speaking about earlier, it's almost entirely dominated by researchers coming from high income countries. So we're at risk of sort of repeating our colonial legacies in the, in the domain of AI. But importantly, the other thing that was really noticeably absent was mentions of how to apply humanitarian principles in this new domain. And so I think that is the real challenge and I would argue there's also a sort of reverse piece. So we know that we can't go on this AI journey just with humanitarian expertise. We know we have to work with the private sector and technology experts. There is a task in terms of capacity building on their side, like for us to strengthen their understanding of the humanitarian sector to help them engage in ways that would be constructive for our humanitarian aid work, and that is on us to help them engage effectively. So yeah, that's our current kind of starting point for thinking about a lot of those governance related aspects. And there's a lot of really exciting initiatives underway as well. And in governance particularly, more than anything else, we need to be making sure we're not just starting up new types of governance mechanisms without coordinating with all of the other good work that's going.
- [37:32]
B
On on Naseem, I'll turn it back over to you in just a second, but there is something that I think that is also important that has not just because of the company that I'm a part of, but because this current funding crisis has brought a lot of other companies like mine together, a lot of the social impact businesses that are really focused on not just the humanitarian side of the house, but also the development side of the house, nonprofits, et cetera, and all of us coming together and starting to learn about what we're all doing because we are as disparate and as competitive, probably more so than even the INGOs are for going for your funding. Right. So we've always been a very competitive small, small group of social impact startups and companies, but there are actually AI companies solely focused on working in the humanitarian sector and working in the nonprofit sector that have all of that stuff. And so I think that there's also a gap. And this is just a point, this is a little bit of grandstanding, so you'll excuse me, but this is a little bit of a gap that I see is that the humanitarian sector engaging with the social impact organizations that are actually there to do good and actually understand the principles and want to work together, that are missing out in the conversations because A, they can't do pro bono work because they're much smaller companies, B, they need funding as well to come forward. And so this creation of partnerships, this creation of co developing of programs together and co seeking of funds together to implement a technology in the right responsible way. And so again, and I think from the GCC perspective, and I know very much so from at least the innovationary perspective, that you guys really are keen on these partnerships. And so as Sean is saying, we need to have these people together. Maybe Zaina. Well, Naseem, I'm going to let you ask the question, but maybe Zaina at some point, before we end the podcast, I'd love to hear your take on partnerships, especially with technology companies or social impact businesses and even the local, you know, local organizations, because the partnership can be just as fruitful. It doesn't have to be with an ingo, et cetera. But Nassim, I don't know if you want to add to that that before Zaina comments. No. Okay, Zaina, go ahead.
- [39:31]
F
I mean, I think at Grand Challenges Canada as well, specifically with Creating Hope and Conflict portfolio. I mean, this idea of partnerships with the private sector is an important theme. We do believe that there is opportunities for humanitarian players and the private sector to tap into each other's resources. And as you rightfully pointed, you know, in the private sector, we have a range of AI solutions and AI expertise that's available to and to the humanitarian sector that we can leverage. I think at Grandchildren's Canada, we're really keen on sort of fostering those partnerships in the same vein that our partners at Innovation Norway are as well, to sort of tap into each other's resources. But I think at the end of the day, it's a partnership, it's a collaboration. I mentioned earlier in the podcast the absence of locally driven solutions. And I think that's a product of like, poor partnership and poor collaboration.
- [40:20]
E
Right.
- [40:20]
F
Like there needs to be not only partnership with the private sector and sort of leveraging their expertise and skill set, but also partnership at the local community level. And I think what we don't see enough is investment in local organizations and collaboration with local organizations. We have a lot of these private sector companies that are trying to drive their agenda, but it really needs to be an adaptation and a collaboration with their local partners, their counterparts part in these effective crises. Because these humanitarian actors are the ones that are coming with that expertise, with that lens of do no harm, with that lens of how can we mitigate risk and how can we be sort of needs, in Theresa's words, needs based and focusing on what is that humanitarian problem that we are trying to solve. And if there's an opportunity for technology and for AI, then great, that's wonderful. But I think there's a role in collaboration where you're sort of weighing these two aspects. I think the main point is we're really trying to ensure that humanitarian actors are leveraging the expertise of the private sector, while also the private sector working with, in collaboration with humanitarian actors to ensure that their product or their service is fit for purpose, is fit for the context. They're mitigating harm and mitigating risk for end user populations.
- [41:38]
C
So speaking about this needs assessment, and I think it's the most critical component around creating the vision or having a vision around what we want to do with AI in our sector. And also a lot of times we see private sectors interested to invest in AI for humanitarians, but the problem statement is not clear. And we've been collecting a lot of data over the years. I don't think we have the problem of lack of data, but I think we don't know what to do with it. But I'm curious, from this group, what are the interesting AI use cases, if you want to call it, or solutions or needs you see, that could be solved or even if not solved, but addressed with AI? Do you talk about this on a daily basis, Therese? Maybe you have a chunk of money set aside for a certain type of solution. What would that be for you? And maybe we can go around the room to discuss this.
- [42:41]
E
Sure. I'd like to go back to the first sort of initial part of your question. When you talk about the needs assessment and the private sector's interest in investing and being a part of a humanitarian response, because we do invest in needs, we don't invest in solutions. So the humanitarian organizations come to us with their need. And sometimes we see that AI becomes a part of the solution. Sometimes AI is the solution, and sometimes it's not, and it's not included at all. But I think one of the things that we want to avoid, that I think we saw, maybe not so much now, but a few years ago, is a humanitarian actor that is working hard in a humanitarian response and being frustrated of the lack of the right tools to help them work efficiently. And they think, okay, I need a new solution. And it used to be an app, now it's maybe AI and my AI solution should do abc. And then they go ahead and they procure that, that. And when they start using it, it doesn't really fit the need. It's not really what they needed. And I asked an AI expert within the health sector, and I said, what do you need from your partner? What do you need from the humanitarian organization side to be able to invest your skill set in humanitarian response? And she said, I need them to know their problem. And so we always start with that needs assessment. And we asked the humanitarian organization or the actors from that side to invest fully and the humanitarian needs assessment, and then we require that they hold an open market dialogue so they present the outcome of that needs assessment. And when we ask AI actors and other private sector actors Are you able to provide your best solutions to humanitarian response today? And if you're not, why not? Then one of the key reasons that come up is a lack of information, a lack of platforms to meet. And so we require that the humanitarian organization hosts a market dialogue where they present the outcome of the needs assessment, they educate their potential partners about what their needs are, and then they listen to the private sector and educate themselves about what are the potentials now, what can we do moving forwards? And based on that dialogue, then they choose the partnership that they want to enter into. And another thing that I just wanted to tap into the whole discussion around partnership is that for these partnerships to work well, they need to be sustainable for both parties. And open source has been mentioned here. And, and I think open source is a really important source of resources for a sector that is extremely constrained of resources. However, if a partnership is not sustainable for both parties, it's not going to last. And if it doesn't last, it's not going to reach its full impact. So we need to figure out how we can establish these partnerships so that they're sustainable for both. And this includes reflections around intellectual property rights. For example, often I think we come into a partnership, we focus our sort of evaluation on what is right on the input, and we say open source, that's right. You know, things should be free and accessible to all. While maybe we should shift that focus a little bit more towards the impact that it can reach. And for a lot of actors with the right skill sets for humanitarian response, intellectual property rights is their income, it's what makes them sustainable. So if they are to be able to engage in humanitarian response, they need to guard their intellectual property rights. And I think there's a lot of options within licensing that can help us bridge the challenges. So making sure that we safeguard humanitarian principles, we safeguard data protection and people affected by crisis at the same time as we allow our partners to be financially sustainable and therefore a sustainable partner for humanitarian actors.
- [46:15]
C
Thank you so much, that was great. Zaina, over to you.
- [46:18]
F
Yeah. I want to jump in on the question of the interesting AI use cases we've seen. I think we've been investing in this, I wouldn't say certainly not exclusively to AI. I mean, we fund innovation across the humanitarian sector, but I mean, naturally AI does feature in our portfolio, at least, particularly in the last few years. And I think there have been some really interesting, fascinating use cases for artificial intelligence. And I think the two that stand out the most for me that I get excited about are the safe optimization tool, for example. And this is an investment that we made several years ago and have continued to support this solution through its transition to scale. And I think that the use case for AI is trying to manage volumes of data, right. And volumes of information. And the SWOT tool has been able to leverage routine water quality data that's already been collected by field teams and applying machine learning and modeling to really generate evidence based and context specific water chlorination targets in displacement settings. So, so it's like really cool use case for how they're taking this volumes of data that's already available and using it to optimize water chlorination in these settings and that will ensure that they can ensure water remains safe up until the point of consumption. And so I just think that that's a really neat use case for the potential of artificial intelligence. And it really, it also speaks to sort of the collaboration and partnership. It was a solution that sort of was started at York University in collaboration with Doctors Without Borders msf. And it's just been a really neat way to sort of bridge the two organizations together and leverage and harness the potential of artificial intelligence. I think another solution that stands out to me in terms of an interesting use case is HOLA Systems. And this is a solution that we also invested in several years ago. And it's an early warning sift to detect incoming airstrikes and provide a 5 to 7 minute warning ahead of an incoming airstrike. And so again, we have a solution here that's been able to leverage machine learning to detect patterns in airstrikes in conflict affected geographies. This solution has gone on to protect hospitals and schools, saving thousands of lives in northwest Syria through this AI driven prediction system. And I think think the use case for AI in this case has been like, we see a lot of predictive tools and forecasting tools. And so like these are two sort of examples that always stand out to me in terms of really neat things that are happening and highly successful, highly useful tools that are really solving an important problem in the humanitarian sector.
- [48:56]
B
I love it. And I actually know both of those systems and I've looked at them both deeply. So I really, I'm glad that you use those two examples because I think they are really strong examples. But I think they also, for example, the halo system, that's the one that gets into that really risk discussion. Right. In our first podcast we were talking about what's your risk level relevant to the error? So if the AI makes an error and doesn't say that something's coming in, you know, and we, we use mine clearing as the example, but this is the same obviously for the missiles coming in the airstrikes. But it's a fine example of what we're trying to do to balance and how we, we need to take the long game with these things and everything else. I don't know if any of you have realized, but we have already reached an hour today on chatting and as I noticed, I think we probably could have done this for two hours today. But you know, we want to respect your time and say thank you to all of you. But as Teresa and Zain, I just got a chance to talk. Shawn, I want to give you the last space to talk about because you talked about that list of 100 use cases that you have at UK HIH that you guys have put out there and what feedback have you gotten from that? What's that creating in terms of the community to be able to discuss these things? Because what you said was again, all the conversation is linked. Teresa was saying open source, you know, potential for collaboration, licensing, etc. You said at the beginning people should look at our list first because then they might see that somebody else has already built it. But are they transferable, the IP and all that stuff? We don't need to get into it, but it is a complex web that we're discussing right around this. So from your perspective, maybe just one use case that you really, you've taken a liking to when you've read what's been done?
- [50:30]
D
Yeah, I mean, I think if I look down that list, there's a couple of things that stand out as the predominant areas of investment at the moment or applications that are being developed. One better or worse is there's quite a lot of flood warning systems being developed. I think they're all probably being developed as the one flood warning system to rule them all. So therein lies a challenge. The other thing that is clear, there's a lot of at the moment is exploratory work on chatbots. And I think it's always good to sense check whether something is a good idea based on one's own experiences to a degree. And as your listeners are absorbing this podcast, I would throw back to you all the times where you may have encountered a chatbot when you're, I don't know, talking to your electricity supplier or, you know, whatever it might be. Most of the time, most of us try and break the chatbot and be like, talk to an operator. And so I only flag that because I'm not saying there isn't a role for chatbots, but I feel like we need to be very conscious of, of what type of service we are trying to provide, what people want, and therefore whether a chatbot is the right solution. I do think before we invest heavily in some of these forward facing, and by that I mean for effective populations to engage in services, those to me are the most high risk areas of AI. And there are actually a lot of quite maybe unsexy but really important back use cases that could help us navigate some of the bureaucracy we have built within the humanitarian system. And for me, some of those are the most exciting opportunities because I remember there was a meme a little while back that I saw that said something to the effect of I don't want AI to create art or write poetry. Those are the things that I want humans to be able to do. I want AI to wash the dishes and do the stuff that is not a great use of my time. And I think therein lies the challenge for the humanitarian sector too, too. Like if we are improving, if we are finding AI uses to alleviate some of the hours and hours that we spend on report writing or sort of pointless backend tasks, how can we then get better at the stuff that's actually really important, like being a good listener to an affected population, acting responsively, making effective decisions. The other area that I think is a major priority for AI that I just want to call into question or help us to think of as a different opportunity is that there are many AI systems designed to synthesize knowledge or to better inform decision making. And while I encourage such efforts, it's very important for us to be aware that there is quite a lot of evidence at the moment for how we could improve humanitarian work. And because we are human, we don't always pay attention to that in decision making or perhaps we don't proportionately give that as much weight as our gut instinct and things like this. So we need to make sure that we are designing, designing and seeking opportunities that will really enhance our core mandate. There is also a chance that because problems are being defined primarily by individual organizations, that we're actually missing opportunities to do some of the bold systems level AI applications that we're not picking up on. Big enough opportunities to create scalable impact across the sector.
- [53:54]
B
Yeah, okay, I will say that that's a whole other podcast, Sean, what you just talked about. So that might be another episode. We might have episode 11 now already already defined. So thank you for that. Yeah, great stuff, Naseem. I think this is it. I think We've reached the end of this amazing conversation today. Nassim was texting me as we were here. She's like, we need to put this out tomorrow. She's like, this has got to go out now. We need people to see this. We need people to hear this. This is a really important conversation. So, Teresa, Sean Zaina, thank you from my side. I really appreciate it.
- [54:26]
C
Naseem, thank you so much. It was a pleasure. Maybe I have confirmation bias because I feel like everything that you were saying resonated with me. So I'm really excited for our listeners to hear, and I think we covered a lot of topics in this one and a half hour, and I think that's the most exciting part about this episode. So thank you so much.
- [54:48]
B
This is what you get when you bring a whole bunch of friends together that love a subject and want to chat to each other and only get to meet like once a year. So this is great. So this was really phenomenal. Thank you all for joining. Thanks for your time today and yeah, can't wait for the next episode. So talk to you soon.
- [55:02]
E
Bye. Bye.
- [55:04]
F
Thanks for having us.
- [55:05]
A
Thank you for joining us on humanitarian frontiers in AI. We hope today's conversation gave you new insights into how AI AI is transforming humanitarian efforts and the steps we need to take to ensure it's 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.