Day 0 Event #189 Toward the Hamburg Declaration on Responsible AI for the SDG

Summary

This discussion focused on the development of the Hamburg Declaration, an initiative aimed at promoting responsible AI for achieving the Sustainable Development Goals (SDGs). The conversation involved representatives from UNDP and the German government, along with various stakeholders. The Hamburg Declaration is part of the annual Hamburg Sustainability Conference, which seeks to bridge the gap between AI and development communities.

Key points included the need to align AI applications with SDG principles while addressing potential risks such as exclusion, environmental impact, and data privacy. The declaration aims to gather voluntary commitments from multiple stakeholders, including governments, private sector, and civil society. Participants emphasized the importance of avoiding duplication with existing AI governance efforts while focusing specifically on development contexts.

The discussion highlighted several crucial areas for consideration, including digital sovereignty, infrastructure development in Global South countries, and the need for AI governance structures in developing nations. Participants stressed the importance of including voices from affected communities and showcasing local initiatives that demonstrate responsible AI use for sustainable development.

Human rights-based approaches were suggested as a framework to address various concerns comprehensively. The organizers welcomed input from diverse perspectives and encouraged ongoing engagement through various channels, including online platforms and future consultations. The process aims to create a pragmatic, action-oriented declaration that aligns efforts to use AI responsibly in support of the SDGs.

Keypoints

Major discussion points:

– Developing a Hamburg Declaration on responsible AI for sustainable development goals (SDGs)

– Aligning AI development and use with SDG principles while addressing risks

– Engaging diverse stakeholders including governments, private sector, civil society in the process

– Focusing on concrete, implementable commitments rather than high-level principles

– Considering environmental sustainability impacts of AI infrastructure and development

Overall purpose:

The goal of the discussion was to introduce and gather input on the process of developing the Hamburg Declaration, which aims to promote responsible use of AI to advance the UN Sustainable Development Goals. The organizers sought to engage diverse stakeholders in shaping the declaration’s content and commitments.

Tone:

The tone was collaborative and open, with the organizers actively seeking input and ideas from participants. There was a sense of enthusiasm about the potential of AI for development, balanced with awareness of risks and challenges. The tone remained constructive throughout, with participants offering suggestions and the organizers expressing appreciation for the feedback.

Speakers

– YU PING CHAN: Moderator

– ROBERT OPP: Chief Digital Officer of UNDP

– NOÉMIE BÜRKL: Head of the Digitalization Unit at the Federal Ministry for Economic Cooperation and Development (BMZ) of Germany

– YASMIN AL-DOURI: Co-founder of the Responsible Technology Hub

– MAI DO: Responsible AI manager at ABUS in Hamburg

– THIAGO MORAES: Works at the Brazilian Data Protection Authority and PhD researcher at University of Brussels

Additional speakers:

– KASSIA: UK delegation to United Nations in New York

– CLAIRE: No role/title mentioned

– TRINE: Representative of the government of Denmark, based in Geneva working on human rights

The Hamburg Declaration on Responsible AI for Sustainable Development Goals

Introduction:

This discussion focused on the development of the Hamburg Declaration, an initiative aimed at promoting responsible AI for achieving the Sustainable Development Goals (SDGs). The conversation involved representatives from UNDP, the German government, and various stakeholders, as part of the annual Hamburg Sustainability Conference’s AI track. The declaration seeks to bridge the gap between AI and development communities while gathering voluntary commitments from multiple stakeholders.

Purpose and Scope:

The primary purpose of the Hamburg Declaration is to address the intersection of AI and sustainable development, focusing on responsible use of AI for development outcomes and embedding this approach in development practice. Robert Opp, Chief Digital Officer of UNDP, emphasised the need to bridge the gap between AI and development communities. Noémie Bürkl, from the German Federal Ministry for Economic Cooperation and Development, stressed that the declaration should align with the Global Digital Compact while being more concrete and action-oriented.

A key point of agreement among speakers was the importance of not duplicating existing AI governance processes. Instead, the declaration aims to fill a specific gap in the AI for SDGs space. The scope includes addressing potential risks such as exclusion, environmental impact, and data privacy while promoting responsible AI use for sustainable development.

Process and Stakeholder Engagement:

The development of the Hamburg Declaration is designed as a voluntary, non-negotiated process open to multiple stakeholders. This approach was supported by both Opp and Bürkl, although they proposed slightly different methods for engagement. Opp emphasised the voluntary nature of the process, while Bürkl focused on utilising existing conferences and bilateral discussions for input.

There was strong agreement on the importance of including voices from developing countries and local innovators. Thiago Moraes, from the Brazilian Data Protection Authority, particularly stressed the need to showcase local sustainable technology initiatives. The organisers expressed their commitment to engaging diverse perspectives through various channels, including an online website for submitting inputs, future consultations, and existing conferences.

Key Issues to Address and Thought-Provoking Comments:

Several crucial areas for consideration were highlighted during the discussion:

1. Environmental Sustainability: Thiago Moraes raised concerns about the environmental impact of AI infrastructure, particularly in developing countries.

2. Access to AI Infrastructure: Yasmin Al-Douri, co-founder of the Responsible Technology Hub, emphasised the need to address access to AI infrastructure and hardware in developing countries, particularly in the Global South.

3. AI Governance Structures: Al-Douri stressed the importance of developing AI governance structures, especially in developing countries.

4. Responsible AI Training and Education: Al-Douri emphasised the importance of responsible AI training and education.

5. Human Rights-Based Approach: A representative from the Danish government suggested incorporating a human rights-based approach to address various concerns comprehensively.

6. Multi-stakeholder Approaches: The need for inclusive, multi-stakeholder approaches that involve those directly affected by SDGs was highlighted.

7. Showcasing Local Initiatives: Moraes suggested bringing in people from the innovation ecosystem and showcasing local sustainable technology initiatives.

8. Balancing AI for Good and Responsible AI: Al-Douri questioned whether the focus was on AI for good or on ensuring AI itself is responsible in achieving the SDGs.

9. Alignment with Global Digital Compact: A representative from the UK delegation asked about aligning the declaration with the Global Digital Compact and clarifying the level of commitments sought.

10. Holistic Approach to SDGs: An audience member emphasised the need for a holistic approach considering potential conflicts between different SDGs.

Implementation and Accountability:

The discussion emphasised the need for concrete, implementable commitments. Opp stressed the importance of balancing ambition with feasibility in making these commitments. Bürkl suggested reviewing progress at future Hamburg Sustainability Conferences. The organisers mentioned the AI-SDG compendium as an opportunity to feature initiatives and track progress.

Conclusion and Next Steps:

The organisers plan to launch a public call for inputs, put the draft declaration online for comments, and present the Hamburg Declaration at IGF 2025 in Norway. Unresolved issues include specific accountability measures and effectively including perspectives from the Global South.

The Hamburg Declaration on Responsible AI for SDGs represents an ambitious effort to promote responsible AI use in sustainable development. While there is general consensus on its importance, the discussion highlighted the need for careful consideration of various perspectives and potential challenges in its implementation.

QR codes were shared for signing up to the email list and accessing the SDG AI companion, providing additional resources for engagement and information.

YU PING CHAN: session, where we’ll explain the process towards the Hamburg Declaration, we’ll also talk about the goals and the aims of the Hamburg Sustainability Conference, and really, this collective effort that we hope all of you will join us in, in really realizing the potential of responsible AI for the Sustainable Development Goals. I’ll start by first calling to the stage, Mr. Robert Opp, Chief Digital Officer of the UNDP, the virtual stage, I think. Robert, please.

ROBERT OPP: Okay. Hello, everyone. This is a strange way of doing a workshop with everyone on headphones. I feel a little weird. I see a few people without, okay, good. All right, just making sure everyone has headsets. Well, thank you, Yuping, and welcome, everyone. On behalf of UNDP, as well as our government of Germany representatives who are online, I wanted to just give a little bit of overview on the way that we as UNDP are seeing the current, I would say, interest and expectations, in a sense, around artificial intelligence, and what we do when it comes to our work to support countries and their national development. And I think it’s fair to say that a lot of us see tremendous potential in AI for supporting the achievement of the SDGs. We already see a proliferation of experimental and sometimes scaling approaches around artificial intelligence and leveraging these technologies in support of different development work, like, for example, in the health space, screening for different kinds of conditions like tuberculosis and other things, helping support small farmers access information on subsidies and other programs that’s available using verbal to text kinds of interaction with chatbots, weather risk modeling, building damage assessment, and so on and so forth. So many examples of application of AI going on right now. But of course, we all know that AI also brings with it a number of risks, whether it be the proliferation of misinformation, or issues like data privacy and protection. But there’s another risk that we also see in the space of applying AI for development. And that is the risk of exclusion, the risk of lack of representation of bias and or inaccuracy in systems, as well as some of the sustainability aspects around the environmental impact of these technologies. So when it comes to the way that we as development actors, and by development actors, I’m referring to international organizations like us, national donor governments, national governments themselves who are implementing their national development programs, civil society actors, and other NGOs, etc. So that whole set of players that are involved in development, we have been looking at how it is that we can improve the alignment that we have around the direction of artificial intelligence application. So what I mean is, if we profess to be working toward the SDGs, we have to be mindful of certain things like the the risk of exclusion, like the potential negative environmental impacts of promoting technologies that consume an enormous amount of energy, for example. And so how do we as a community come together and really align ourselves and work toward a sort of set of directions or a set of commitments that we can make together as a community moving forward. So the Hamburg Sustainability Conference was an event that was sponsored last October by the government of Germany, and featured a lot of discussions around the practice of sustainable development. and the future directions of sustainable development. And my colleague from BMZ in Germany will be speaking in just a second and she’ll also address the kind of background of the conference and things like that. But in the Hamburg Sustainability Conference, there was a track that was set aside for AI and digital. And it really looked at the different aspects of responsible application of artificial intelligence and digitalization in the development space. And we looked at a number of things, including specifically the environmental impacts and some of the other aspects. But we also took the opportunity to start convening a panel that was focused on the principles that we wanna work toward for alignment around using AI for development. And we focused those principles around the five P’s, the five principles that are as part of the agenda 2030. So people, prosperity, planet, peace and partnerships. But now, so we had a very good discussion. There was a lot of interest from stakeholders. Generally in Hamburg last year, there was a high level of participation and a very multi-stakeholder participation as well. And now we are moving forward to thinking what is it that we can move toward in next year’s Hamburg Sustainability Conference in this space of artificial intelligence and the SDGs. And we want to continue convening these discussions around what are the areas of commitments we can make, as well as how are we collecting and gathering information on what’s out there. And to that end, one thing I just forgot to mention is we launched at last year’s Hamburg Sustainability Conference, an SDG compendium or AI compendium that starts to gather AI examples that have been used in the development space. In other words, trying to pull together a sense of where the practice is out there as well. So final thing I would say before turning back to you Ping is we do see all of this as a direct part of the follow-up to the Global Digital Compact. Paragraphs 53 and 54 of the Global Digital Compact talk about the application of AI to the sustainable development goals and the promise that they put forward. And so this effort is really seen as taking the next step in terms of collective commitments we can make that are more granular and instead of high level and saying, what are the things we can align around to really work together as a global community when it comes to pursuing the SDGs with artificial intelligence. With that, I’ll turn back to you Ping.

YU PING CHAN: Thank you, Rob. And as Robert said, we’re really looking to this session to be a little bit more of an engagement with you to really think through what we could have as part of the work towards responsible AI for the SDGs. So not just for the Hamburg Declaration, but also around convening these types of discussions at the Hamburg Sustainability Conference, which as Rob says, will be a unique opportunity to make sure that development practitioners which are gathering at Hamburg will also take into account what perhaps technologists and internet governance experts such as yourself convening here for technology conference really need to keep in mind. How do we bring these two communities together? So now, before we really go to hearing from you, we would like to invite our second speaker who is online, Ms. Naomi Burke, who is head of the Digitalization Unit at the Federal Ministry for Economic Cooperation and Development, the BMZ of Germany to speak. Naomi, please.

NOÉMIE BÜRKL: Yes, thank you so much. Thank you. you Ping and Rob, and welcome to all of you from Germany. It’s a pity I can’t be with you today, but the technical tools we have today allows me to be with you, so let’s use them like that. A lot of what I wanted to mention has already been said, and for the sake of time, I don’t want to to repeat them, but also to confirm that from a German government perspective, we do see artificial intelligence as a driver for achieving SDGs or for at least accelerating the implementation of SDGs there. The sectors that we could look at, agriculture, shows that AI can enable analysis for climate and crop data to adapt to climate change more effectively. Health has been already mentioned by Rob, where AI can distribute health information during epidemics, for example, and education, of course, we see that AI can help personalize learning. This is why the Ministry of Development Corporation has been engaged since 2019, actually, as a partner to support also the use of AI in this respect. It has potential, but also the risks that have already been mentioned, talking about, for example, water and electricity consumption, discrimination or disinformation. This is why we want to really focus on how to use AI in a responsible way to ensure that AI serves people and planet. On the HSC, the Hamburg Sustainability Conference, maybe just to say very broadly. It is an initiative that facilitates an exchange based on mutual trust and partnership between leading international minds from politics, international organizations, private sector, academia, civil society on those structural issues that we see. And this is why I think it is very good to know it is not just a one-off conference, it is an all-year and multi-year process. And we really want to take the time to discuss with you what we can do to underscore this need of commitment that has been mentioned, linking also to implementing the efforts underlined in the Global Digital Compact as well. So yes, the SDG Compendium has already been mentioned. A warm welcome to you to participate in that, to look at that, to contribute into that process. We have discussed principles to see what we mean when we talk about a responsible use of AI. And we really want this to be an inclusive and collaborative effort. And so thank you very much UNDP to convene all the minds that can contribute to that. And I’m really looking forward to your support, to your engagement and to using AI in a responsible way to have a boost for the SDGs. Thank you.

YU PING CHAN: Thank you so much, Naomi. And I really want to welcome new colleagues that just came into the room and say that we’re looking forward to having an engagement with you, not just through this particular meeting, but also throughout the entire process towards the Hamburg Declaration. And to that end, I’ve actually circulated some documents, a copy of the, some of the background around the SDGs. So if you have any questions, please feel free to reach out to me. the Hamburg Sustainability Conference that encapsulates what Rob and Naomi had just briefed but we’re also happy to provide more copies. I’ve also sent around an email list and so if you could leave your email addresses on it if you’d like to stay and engage with this process as we develop the Hamburg Declaration. We also have a couple of links online that I think I’ve put there but if not I’m happy to repeat the links later so that you can sign up online as well. So basically starting with this workshop and moving towards the next Hamburg Sustainability Conference which will be in early June 2025 we will be convening a number of these both online as well as in-person consultations on the content of the Hamburg Declaration. It will be as Rob has explained really thinking about what we together as the global community can come together to think about what are commitments or areas of action that we think need to be committed to or agreed on so that we can realize responsible AI for the SDGs and it’s really a very iterative process. We don’t have a draft in mind, we don’t really have areas that we want to focus on beyond the guiding principles of the Agenda 2030 and so it really would be shaped by your contributions and inputs as well. We’ll also have an online website where you can submit such inputs in writing if you or your organization would like to contribute something towards the thinking process as well as possibly even convene consultations of your own as well. So we’ll have some background material that you can use to also convene these types of informal conversations around the content of responsible AI for the SDGs. We also want to emphasize it’s not really a one-off, right? It’s not that we would necessarily come up with something at the Hamburg Conference in June of next year and end there because Hamburg as Nomi has explained will be an annual meeting. There will be an opportunity to continuously reflect back on these areas of responsible AI for the SDGs. We do think that it will be a start of a continuing conversation that will be multi-stakeholder in nature, hoping to bring in these commitments not just from the private sector but also donor institutions, governments and development actors as well towards AI for the SDGs. So having sort of started with that point, I think what I’ll do first is maybe open up the floor to any questions that colleagues might have. around Hamburg, the background, the Declaration, before we dive right into the content. Would there be any questions both online and offline? I also want to welcome, I think there are about 20 colleagues online as well. I see a couple of questions in there about the content of the Declaration and maybe some of the specific areas that we had discussed, so I’ll leave those for a little bit later, but I’ll just start with any questions around the process, the Hamburg Sustainability Conference that happened just this October, before we open up into the substance itself. Ah, please, come to the standing mic and introduce yourself.

YASMIN AL-DOURI: Okay, now I can hear myself as well. Hi, my name is Asmina Alduri, I’m the co-founder of the Responsible Technology Hub and I have a very maybe general question. When we talk about responsible AI for the SDGs, are we talking about AI for good? Are we talking about AI that is used to get to the SDGs or are we talking about AI that needs to be responsible to get to the SDGs? They are two different things. So this is like what I was wondering earlier also in the talks.

ROBERT OPP: Noemi might have a take on this as well, but if I’m understanding where you’re coming from, this is about responsible use of AI, which for me includes ensuring that AI systems themselves are responsible. So we’ll try to, because this is the distinction I think we’re making between, you could utilize AI for a development end or output of some kind, but if you’re not, let’s take a concrete example. Let’s say that we think that AI could revolutionize the education platforms in certain countries and so we invest a lot in creating a lot of compute power and extending AI systems to students, etc. But if we’re not mindful of the sustainability footprint of those AI systems, we’re actually creating another problem while we’re trying to fix one. And so it’s actually about both of those. It’s about utilizing AI for development outcomes, but doing that in a responsible way. So I hope that answers the question.

YU PING CHAN: Noemi, if you might want to come in on that as well.

NOÉMIE BÜRKL: No, I would support that. And… I would add the dimension of, but this is actually what you just said, Rob, that we need to make sure that when we do believe that AI can actually contribute to a positive development outcome in one area, we have to make sure that we also see at the same time the potential risks and be mindful of that as well. I mean, HSC, why we bring this topic in, to also maybe explain a little bit overall, it is not per se an AI conference. It is a conference on how to promote the achievement of the SDGs by 2030, which is a very difficult task. And this is where we see the role of AI and digital, both in the positive and negative sense. So you can say AI for good in a way, but yes, I think it is more holistic than that. Thank you. Always focused on development aspects though.

AUDIENCE: From the UK, Kassia? Hi, colleagues. UK delegation to United Nations in New York. From the perspective of GDC negotiator, have you thought about aligning the lines of declaration with GDC more like clearly, just for consistency? So I’m just asking this question to maybe stir the pot a bit. And the second question is, do you know what the level of commitments that you want to achieve in the end? What’s the ultimate goal in terms of commitment?

YU PING CHAN: Pardon the passing of mics over here. Before I give the floor back to Rob and Naomi to respond to Kassia’s question, I think this is linked to another question that I’m seeing in the chat from Monica, which is, how is the Hamburg annual conference linked to GDC AI follow-up processes? Does it stand apart? And how is it linked to the IGF dynamic coalition on data and AI? I think very quickly before I turn it over to them, I would say we would very much welcome a link between the IGF dynamic coalition on data and AI, if the dynamic coalition wanted to think about possible inputs that they could actually submit towards this process, as well as to perhaps, we would very much value the network and the coalition being part of the consultation forward. So over to Rob and Naomi.

ROBERT OPP: Okay. Well, thanks for the questions on the link with the GDC, and I was just pulling up my copy of the GDC right now. If I understand your question correctly, so we see this process as contributing overall to the, I would say, implementation of the GDC or in the spirit of the GDC. For sure, I expect elements that are mentioned in the GDC to come up in this process. So GDC talks about the importance of capacity building, It talks about like a lot of other aspects around technology in general and then the AI pieces themselves. The. But in terms of, like, direct linkages with for example the, the scientific panel that is proposed around AI and some of the things as well we don’t know yet because those mechanisms are not in place. I suspect that there could eventually be a link. There’s also the proposal in the GDC around the global annual dialogue on AI governance. Again, that may form a part, but we’re not trying to address the issue of international AI governance with this process. This is much more as Noemi also just reminded. What we see is there’s a whole set of summits going on globally around AI safety and AI governance and so on and so forth and there will be another one there was Bletchley Park. There was one in Korea, there will be another one in Paris, there will be a galley talking about AI in Africa. In April, and so on and so forth. The Hamburg Sustainability Conference is not trying to be one of those. The Hamburg Sustainability Conference is a conference on development, which has an AI track. And so we’re trying to ensure that we don’t just keep this discussion of AI and technology in those AI and technology focused conferences but actually we’re embedding it in our development practice, because to be frank, that’s where the big money flows when it comes to overseas development assistance and many other forms of bilateral and multilateral cooperation. And so we want to be sure that the practice of development is infused with this responsible AI utilization and application for SDGs. So I hope that answers today.

NOÉMIE BÜRKL: I think Rob said it very nicely and maybe also because I just saw another question on the chat. Why do we need another declaration? We want to make sure that there is no duplication to other processes. When we came up on these issues, really what stands out for us is we have on the one side and this was just just mentioned we have the SDG community discussing and we have the AI community on the other side discussing. And what we really want to focus on is on the implementation aspects of those paragraphs on the GDC that that that were mentioned. But also to go a step further, because what we do bring in together here are the players from the private sector, from academia, civil society, et cetera, and to really continue what we see in the agency is that we do not have that formalized discussions, which is good because we want to move forward on really this particular issue and see how how far it takes us there. And I think that some actors are really quite willing to participate in this process and we really need them as well. So, yes, we are mindful of not to be duplicating other aspects.

ROBERT OPP: Noemi, you just also reminded me of a key aspect of this declaration, which is that it is not intended to be a negotiated process, meaning we’re not trying to get universal adoption here as such. We hope that everyone will come and commit to it, but it’s going to be a voluntary kind of thing, not a negotiated process like the GDC or some of these other intergovernmental or universal kinds of agreements.

YU PING CHAN: And again, on that particular point, even though the AI space is very crowded and the UN space is very crowded, we do think that there is a gap when it comes to this idea of AI for the SDG. and particularly from a development perspective. So that is the gap that we’re looking to fill. And as the United Nations Development Agencies practicing in this field, we have noticed that there needs to be that coming together of these types of communities. I saw a comment or a question over there, please.

MAI DO: Hello, can you hear me? Yeah, thank you. So, Mai Do, I’m a responsible AI manager at ABUS in Hamburg. So I work for the civil part of ABUS. And I was very interested about the aspect of the involvement of the different stakeholders. And as you mentioned before, it’s a voluntary commitment, right? So, therefore, how do you envision to bring those different stakeholders together from the private sector to the civil society to ensure building a resilient infrastructure, which is one of the goals you’re trying to promote? So how do you envision it? And how do you go about it? And lastly, how do you think of holding these people or the different stakeholders accountable to ensure these goals? Thank you.

YU PING CHAN: Thank you for the question. I think maybe given that this is a German stakeholder and you’re coming from the perspective of Hamburg, I could reverse the order and ask if Naomi could take that question first, followed by Rob.

NOÉMIE BÜRKL: Well, thank you very much. What we really want to do is, we’re kicking off the process, right? So we really want to go with those who feel that they can have and really make a difference in that. So we do want to involve all those interested. And in this room, we have those minds that I mentioned before, and also online. And I think that’s really important. think that is exactly what we’re trying to aim for. It will be a process where we will also use those conferences mentioned before in Kigali and in Paris, et cetera, to also involve all those partners. But we will also have discussions on a bilateral basis, for example, with firms that have already shown their interest, SAP, et cetera, and others in the private sector. Because we really think that there is a huge potential there. And of course, civil society and academia. I know that being involved in those GDC discussions, there may be some hesitancy in why we need this. But we really want to become more concrete and in terms of how we can really have a major boost in these SDGs. This is a very, actually, narrow approach to that. And we will not be able to tackle all the SDGs in all the areas. We are looking at five principles. We will be also looking at the ideas that we have. And we hope to be as ambitious as we can. But we will have to see how the process goes in the coming months to see what we can agree on. And on this, after that, it will depend how we will make sure in terms of accountability. We will keep looking at the next HSEs after that. Looking back, how far did we go with these commitments so that it is beyond mere agenda setting? And what comes out of that? We do believe that those who are part of the process will also be those who are also convinced and willing to participate. So I’m still very hopeful that this is a very good approach to go. And also, we are. I want to underline again, we want to be very concrete as well. Thank you.

ROBERT OPP: I don’t have a lot to add to that except to say in the Hamburg Sustainability Conference, I’m not sure if you were there or not, but on the discussion that kicked all of this off, we had the head of sustainability from SAP there and inside discussions with him, he was saying this would be of interest to his company to align if those principles and the eventual commitments can make sense and so on. And we’re still kind of designing what this actually looks like. We very much welcome private sector to sign on to the commitments as well. And there will likely be a number of ways to do that. So more to come on that.

YU PING CHAN: And so moving on to really that concrete part of it, if we could move on maybe from the questions around process, do we welcome them subsequently as well, and maybe get into the meat of what the declaration should look like. And here again, we’re just really looking for ideas and inputs as to what you think are critical issues in the AI for SDG space, or as maybe a development practitioner, what do you think when you think about AI? What does it mean to use AI responsibly? And conversely, from a technologist or AI scientist space, if you’re looking at how AI is being used in development space right now, what are your concerns? What are your thoughts on the risk? What do you see as the opportunities? And so let’s have this as a little bit more of that open discussion that Rob was speaking to a little bit earlier on, in this thought that if you could really have curate a conversation or create an opportunity for the private sector, the multi stakeholders, the governments to come together around these issues, what do you think should be top of mind for them? Anyone in the room would like to take a first stab at this? If not, while you are ruminating on that, I actually have a question from the online chat already, and I will direct this at both Rob and Naomi, where? So the question here is about the, whether the Hamburg Declaration will be using. we’ll be dealing with issues around artificial intelligence-based weapons, where there has been a concern over the use of such weapons in the ongoing situation in Gaza. And so what could be the role of the UNIGF and the United Nations in the fight against the weaponization of artificial intelligence? And so I would turn that over to Rob and Naomi.

ROBERT OPP: Okay, so the issue of weaponization of AI will not be tackled directly by this process, we don’t foresee. There are elements of the peace principle that we need to respect in terms of the way that AI is applied so that it doesn’t promote divisions among people and so on. But because there are other parts of the United Nations multilateral system, that like disarmament affairs and things like that, that are dealing with some of the weaponization issues, we don’t feel that this is the best process placed to actually do that. So I think we need to acknowledge it in some way, but likely not be, this is not meant to be the platform to really address those issues which are being taken up elsewhere.

NOÉMIE BÜRKL: Yes, I agree, absolutely. Because I think there is already so much to do on these issues that I mentioned before. Also, people, prosperity planet, etc. We don’t want to duplicate processes that are being discussed elsewhere. And I think there’s another question also on presenting then the declaration at the IGF 2025 in Norway. I don’t see why we shouldn’t do that. That’s also interesting to look at, because we really want to link the different processes. But I think it’s good to get all the ideas here in this room. and to shape the process together. This is what this is really about.

YU PING CHAN: Naomi, can I also ask you to answer the second part of Dennis’ question online in the chat, which is if the conference is by invitation only?

NOÉMIE BÜRKL: Yes, it is by invitation, but we of course have a say in who is invited. I think what we will do is to look at the process, who is involved and who would like to participate. I think we can look at that flexibly in the coming months.

YU PING CHAN: Any other questions from here? Yes. Please introduce yourself.

THIAGO MORAES: Thanks. My name is Thiago Moraes. I work at the Brazilian Data Protection Authority and also as a PhD researcher at University of Brussels. Responsible innovation in AI or other emerging technologies has been part of my research topic. The more I look into that, I see sometimes a bit of a dissonance on how we as governments, using a bit of the government hat, we’re discussing a lot these days about digital sovereignty and how we have to raise the capacity of infrastructure, especially in global majority countries that usually have more challenges for that. We just participated this year as the host of Digital Training Brazil. The Digital Economy Working Group was discussing a lot about this importance of raising the capacity level. But then, and this here now is more of my academic hat that I want to bring, it’s like I think as well that I miss a bit of the part of what I’m doing. I miss a bit of the part of what I’m doing. I miss a bit of the part of actually is being doing for the sustainable part because when we are so concerned about digital sovereignty and actually creating more infrastructure for better data centers, better you know like just having more data power if we don’t think of the other side of the balance and how we are actually promoting that and for sure green energy like green data centers is that concept really exists is definitely part of it but even like environmental impacts because it’s not only about using you know like okay Brazil use a lot of energy from water sources which supposed to be cleaner for sure but still there’s a lot of environmental impacts that sometimes we create and this should be part of the discussion so if several countries now are working to build better data centers more represents and they don’t add this to the equation we’ll have a lot of trouble in the upcoming years so maybe I know it’s a voluntary declaration but this somehow should be embedded there in the discussion I think that’s my suggestion. Thank you and I think that’s an important suggestion and

YU PING CHAN: we will definitely look to taking that up under the planet part of the declaration but really indeed as you say this environmental sustainability and bilingual as we look to build out compute data and the AI revolution is really critical. Responses, Naomi, Rob.

ROBERT OPP: No just to say absolutely yes and the one of the challenges we are going to need to address over the next few months as we do this is what can those commitments look like and so that’s also why we welcome the participation of many voices to help us understand what would actually be feasible and implementable ways of putting those kinds of commitments in because I don’t think it’s quite as simple as just saying as you mentioned okay it’s data centers have to be carbon neutral or something right like it’s there are other aspects of the issue that we need to explore and then eventually balance the perfect with the feasible of what can actually be implemented what people can commit to so just completely agree and since you’re a PhD researcher and government authority we would definitely welcome both sides on the challenge of academia with the kind of implementation necessity or need or how what governments would be able to do.

YU PING CHAN: Yes there was a question over there and then I think another one over here.

AUDIENCE: So I hope yeah it’s working good. This is less of a question and more of an answer to your question that you had. So I basically work in the field of responsible AI used to work for it for big corporations big tech in the past before I was one of the co-founders of the Responsible Tech Hub and then Munich where we focus on these topics specifically from a youth perspective and there are a couple of things that I think or I deem as super important when we talk about SDGs and using AI to harness it the first thing is what SDGs are you focusing on after you actually really define which SDGs you focus on you can actually go into the aspect of okay who has access to the infrastructure and who has access to the hardware. As long as these questions are not answered there’s no way we can even include for example the global south or anything sub-saharan if we focus on Africa. That’s the one thing. The other thing is also if we talk about responsible AI training is the number one aspect that we’re not only focusing on in Germany right now but generally in the EU. So yes we have the EU AI Act for example but there is a lot of governance still lacking. I was just at a session where we discussed AI governance structures for the Middle East that is barely even existing so there’s still a lot of room to talk about AI governance in different countries specifically in developing countries so this has to be set as well and then there needs to be training for those who are actually developing the AI and who are actually deploying it. I think there’s a lot of resources out there. There are a lot of institute the Alan Turing Institute the Tom think tank which I also represent in some kind of ways. There are a lot of academic resources to go into AI impact assessments for example. So that already exists but it’s super important to keep in mind that if we talk about SDGs we always have to include those who are directly affected by the SDGs and those are mostly the ones who don’t have the access AI and to actually access the training and to have the base, which is AI governance. So these multi-stakeholder approaches have to actually happen first, I believe, before we can even set up trainings for them. Comments in the room. These are incredibly helpful. Hi, I’m Claire, and I wanted to ask a question which is somewhat in the similar direction, and that is that the point I found most interesting is that point of conflict between SDGs, maybe. And I was wondering, because if I look at the document, you ask for input for special areas. So I don’t see the representation of somewhat of a general point of looking at, well, if I want to promote a certain point, I also have to incorporate others, as well as, especially if you look at AI and that most of the use cases are based on the data, obviously. And I mean, that is probably the area where we are lacking most in responsibility and in humanity, and whether this will also be part of it to look at the development point from a more holistic point of view and incorporating those SDGs, as well.

YU PING CHAN: I think that’s a great point. And yes, for convenience sake, we did sort of split into those five Ps, but we do expect there will have to be some kind of chapeau, as you say, like looking at AI more generally and then maybe touching on data that would then fit across all of this. So thank you for that. I also want to say, really, we appreciate those of you that are really looking at this from your practitioner, but also expert perspective. So if you could make sure to pass me your contact details and card or sign up online to the website and the email list, we really want to stay in touch with you and have you as part of the process. Rob, Nomi, while there are any other questions in the room, over here, please. I think after this and another comment, we’ll turn it back to panellists very quickly and come back to comments.

AUDIENCE: Hi, and apologies in advance. I’m not quite sure if this will turn into a comment or a question. My name is Trine, I’m a representative of the government of Denmark, but I’m normally based in Geneva working on human rights. So I’m actually neither a development or an AI practitioner. But I got inspired by your talk about not duplicating, not reinventing the wheels, but actually making sure there’s complementarity. And of course, in the human rights field, we also work a lot on on AI and the negative, mostly consequences of it. And I think a human rights based approach to whatever you are doing, be it AI or the SDGs, actually, is very helpful in that sense, because if you make sure that you have a human rights based approach, you don’t only cover those very obvious elements such as discrimination, AI bias, all of that, if you indeed incorporate it in the design, development and deployment phases. So when you talk to those actors, when they develop the products, basically, and I think that’s also partly where the education comes in, those communities talking to each other. But you also have the right to health. You have the right to a clean, healthy and sustainable environment. So indeed, the human rights framework is not only respected and accepted by all states, but it’s actually also very well developed and well versed. So I think it probably turned into a comment that that would be something to look at. Thanks.

YU PING CHAN: More comments in the room or online from online colleagues and participants. Again, it doesn’t have to be focused, per se, on the Haber process. as a declaration, maybe just what do you worry the most about when you look at the use of AI today? Particularly interested also in perspectives from developing countries and the global majority. A follow up.

THIAGO MORAES: Yeah, OK. I mean, it’s always nicer when we have more perspective. But just in addition, another thing that I think it could be really interesting for any conference, not only the number one, but when we want to have more concrete results. I know, of course, policy-oriented conference, in the end, you try to end up with a statement, but also, I don’t know, I mean, it’s the first time I’m hearing about the conference, so maybe you already do that. But one thing that I miss in conference in general, like the IJF, for example, is have more of like showcases of local initiatives that are actually making this kind of difference. So for example, let’s say, why not bring some people that are actually bringing a smart, clean, sustainable way of using a certain type of technology and reaching, even if just in a local level, what’s happening in some places. Also bringing a bit more of people from the innovation ecosystem. And nowadays, we see a lot of different initiatives like the sandboxes, innovation hubs, and experimentation facilities that are also trying to bring more of this discussion of responsibility in AI development, or sustainability, et cetera. And usually there are some small use cases, and maybe for them to become scalable, we have to look more into them. So if this is not already done, maybe it’s something that could be nice to have during the conference, you know?

YU PING CHAN: We think that’s a great idea. We look to our German colleagues who are actually supporting more the organization of the conference. I would also say, I think we had mentioned that we had launched the AI-SDG compendium, where there could be an opportunity to also feature these initiatives. And from the UNDP perspective, because we are present in so many countries and really are looking to globally scalable, as Sergio said, initiatives in this area that really represent the ability to affect across, please don’t hesitate to reach out to us as well.

ROBERT OPP: Yeah, maybe I can just add to that by saying we have a, you know, as Yu-Ping said, countries in all of the country, representation in all countries across Africa and many others, so 170 countries in total. Many of those places, we actually are tied into the local innovation ecosystem. And there’s been a couple of incidents recently, where we have been forming networks of African innovators in particular, and featuring their participation in global conferences and discussions, which is very interesting, because you need those voices around the table. Also in the Hamburg conference last year, we brought an indigenous activist from Chad to discuss the aspects from her perspective, what is actually happening on the ground, what is touching people there, or indeed not representing them well when it comes to the rollout of technology and AI. So this is absolutely, just to say, completely agree with your point. And the conferences are very, especially those that are based in the global north, often have trouble getting the representation that we need to have the right discussion around the table. But that’s also another reason why the IGF and some of these multi-stakeholder forums are important to have as part of our consultation process.

YU PING CHAN: We’re running out of time. I see a reminder. I just really want to give an opportunity to anybody who still has any comments or suggestions. Again, recognizing this is the first time that we are opening this up, so again, we welcome future contributions. If any of the dynamic coalitions or the youth or regional IGFs want to take up the Hamburg Declaration or even the conversation around responsible AI for the SDGs, please let us know. Reach out to us. We’d be more than delighted to do that. Any last comments from those online, offline, here in the room? Comments, observations, and so on. I also ask my colleague Marie to share a QR code for you to scan to stay updated with us at the very end. So maybe Nomi first, then Rob. Nomi?

NOÉMIE BÜRKL: Well, thank you very much to all of you for your good questions, good ideas. This is exactly what we were going for. We hope you will be or stay engaged. We’re very excited about the future months to come. Please do also use the online possibilities to reach out, as Yuping just mentioned, and yes, looking forward to your ideas. Thank you.

ROBERT OPP: Similarly, from my side, just to thank you for participation. Thank you for the comments. It is exactly what we had hoped that we would get out of this session. It is a good reminder about the kind of importance of collective brain power. So we want to have something good. We want to pressure test it from a lot of angles. It’s not going to be perfect. We know that from the beginning. But what we want is something pragmatic because the practice is evolving so quickly out there that we want to try to stay ahead as much as we can and start to align our actions, our commitments, so that we really are making sure that AI is used in the right direction and for the actual support of people, putting people and their rights at the centre. So that’s just a thanks from our side for all these good ideas and comments.

YU PING CHAN: Shear screen very quickly. Marie, you should have permission that these are the QR codes for you to be able to sign up to the email list. I think, Marie, you need to full screen. There we go. So that’s www.bmz-digital.global.en. And then here we also have the SDG AI companion that I mentioned, where we also do welcome initiatives, especially at the global level, where you feel that it fits this idea of responsible AI for the SDGs. And we really look forward to keeping in touch with all of you and really taking into account your views and perspectives. We’ll also have a public call for inputs later on. We’ll also put up the declaration online so that we can take comments and so on. And really looking forward to this being an engaging process and thinking about how the IGF can contribute not just here, but also into other global processes as well. We look forward to working with all of you. And thank you again for sharing your time with us today. 笒笒笒笒笒笒笒笒笒笒笒笒笒笒笒笒笒笒笒笒笒笒笒

Agreement Points

Focusing on responsible use of AI for development outcomes

ROBERT OPP

NOÉMIE BÜRKL

Focusing on responsible use of AI for development outcomes

Aligning with the Global Digital Compact while being more concrete

Both speakers emphasize the importance of using AI responsibly for development outcomes, aligning with existing frameworks while providing more concrete actions.

Not duplicating existing AI governance processes

ROBERT OPP

NOÉMIE BÜRKL

Voluntary, non-negotiated process open to multiple stakeholders

Not duplicating other AI governance processes

Both speakers stress that the Hamburg Declaration is not intended to duplicate existing AI governance processes, but rather to fill a gap in the AI for SDGs space.

Importance of multi-stakeholder engagement

ROBERT OPP

NOÉMIE BÜRKL

THIAGO MORAES

Voluntary, non-negotiated process open to multiple stakeholders

Utilizing existing conferences and bilateral discussions for input

Including voices from developing countries and local innovators

All three speakers emphasize the importance of engaging various stakeholders in the process of developing the Hamburg Declaration.

Similar Viewpoints

Both speakers highlight the importance of considering infrastructure issues in developing countries, including environmental sustainability and access to AI hardware.

THIAGO MORAES

YASMIN AL-DOURI

Environmental sustainability of AI infrastructure

Access to AI infrastructure and hardware in developing countries

Both speakers emphasize the need for concrete, implementable commitments and ongoing review of progress in future conferences.

ROBERT OPP

NOÉMIE BÜRKL

Making concrete, implementable commitments

Reviewing progress at future Hamburg Sustainability Conferences

Unexpected Consensus

Importance of local initiatives and voices

THIAGO MORAES

ROBERT OPP

Showcasing local sustainable technology initiatives

Voluntary, non-negotiated process open to multiple stakeholders

Despite coming from different perspectives, both speakers unexpectedly agree on the importance of including local initiatives and voices in the process, particularly from developing countries.

Overall Assessment

Summary

The main areas of agreement include the need for responsible AI use in development, avoiding duplication of existing processes, multi-stakeholder engagement, and the importance of concrete, implementable commitments.

Consensus level

There is a moderate to high level of consensus among the speakers on the overall approach and key principles of the Hamburg Declaration. This consensus suggests a strong foundation for developing a meaningful and impactful declaration on responsible AI for SDGs. However, there are still areas where more detailed discussions and alignment may be needed, particularly on specific implementation strategies and addressing the unique challenges faced by developing countries.

Different Viewpoints

Scope of the Hamburg Declaration

ROBERT OPP

NOÉMIE BÜRKL

Addressing the gap between AI and development communities

Aligning with the Global Digital Compact while being more concrete

While both speakers agree on the need for the Hamburg Declaration, they emphasize different aspects of its scope. Opp focuses on bridging the gap between AI and development communities, while Bürkl stresses alignment with the Global Digital Compact and providing more concrete actions.

Unexpected Differences

Environmental sustainability of AI infrastructure

THIAGO MORAES

ROBERT OPP

NOÉMIE BÜRKL

Environmental sustainability of AI infrastructure

Focusing on responsible use of AI for development outcomes

Not duplicating other AI governance processes

While Moraes raises concerns about the environmental impact of AI infrastructure, this issue is not explicitly addressed by Opp or Bürkl in their main arguments. This unexpected difference highlights a potential gap in the current focus of the Hamburg Declaration.

Overall Assessment

summary

The main areas of disagreement revolve around the specific focus and scope of the Hamburg Declaration, the methods of stakeholder engagement, and the extent to which environmental sustainability should be addressed.

difference_level

The level of disagreement among the speakers is relatively low, with most differences being more about emphasis and approach rather than fundamental disagreements. This suggests that there is a general consensus on the importance of responsible AI for SDGs, but some refinement may be needed in defining the specific goals and methods of the Hamburg Declaration.

Partial Agreements

Both speakers agree on the importance of stakeholder engagement, but they propose different approaches. Opp emphasizes a voluntary, non-negotiated process, while Bürkl focuses on utilizing existing conferences and bilateral discussions for input.

ROBERT OPP

NOÉMIE BÜRKL

Voluntary, non-negotiated process open to multiple stakeholders

Utilizing existing conferences and bilateral discussions for input

Similar Viewpoints

Both speakers highlight the importance of considering infrastructure issues in developing countries, including environmental sustainability and access to AI hardware.

THIAGO MORAES

YASMIN AL-DOURI

Environmental sustainability of AI infrastructure

Access to AI infrastructure and hardware in developing countries

Both speakers emphasize the need for concrete, implementable commitments and ongoing review of progress in future conferences.

ROBERT OPP

NOÉMIE BÜRKL

Making concrete, implementable commitments

Reviewing progress at future Hamburg Sustainability Conferences

Key Takeaways

The Hamburg Declaration aims to address the gap between AI and development communities, focusing on responsible use of AI for SDGs

It will be a voluntary, non-negotiated process open to multiple stakeholders

The declaration will align with the Global Digital Compact while being more concrete and implementation-focused

Key issues to address include environmental sustainability of AI, access to AI infrastructure in developing countries, and AI governance structures

The process will incorporate voices from developing countries and showcase local sustainable technology initiatives

Resolutions and Action Items

Launch a public call for inputs on the declaration

Put the draft declaration online for comments

Utilize existing conferences and bilateral discussions to gather input

Create an AI-SDG compendium to feature initiatives and use cases

Present the Hamburg Declaration at IGF 2025 in Norway

Unresolved Issues

Specific commitments and accountability measures for stakeholders

How to balance digital sovereignty with sustainability concerns

Extent of addressing AI weaponization within the declaration

How to effectively include perspectives from the Global South in the process

Suggested Compromises

Balance ambition with feasibility in making commitments

Incorporate human rights-based approaches while focusing on development outcomes

Address environmental sustainability without duplicating other AI governance processes

When we talk about responsible AI for the SDGs, are we talking about AI for good? Are we talking about AI that is used to get to the SDGs or are we talking about AI that needs to be responsible to get to the SDGs? They are two different things.

speaker

Yasmin Al-Douri

reason

This question cuts to the heart of defining the scope and goals of the initiative, highlighting an important distinction between using AI as a tool for development versus ensuring AI itself is developed responsibly.

impact

It prompted clarification from the organizers that the initiative aims to address both aspects – using AI for development outcomes while ensuring the AI systems themselves are responsible and sustainable. This helped frame the subsequent discussion.

From the perspective of GDC negotiator, have you thought about aligning the lines of declaration with GDC more like clearly, just for consistency? … And the second question is, do you know what the level of commitments that you want to achieve in the end? What’s the ultimate goal in terms of commitment?

speaker

Kassia (UK delegation)

reason

These questions probe important aspects of how the Hamburg initiative relates to existing processes and what concrete outcomes it aims to achieve.

impact

It led to clarification that the Hamburg process is meant to complement rather than duplicate other AI governance efforts, focusing specifically on development applications. It also highlighted that the declaration will be voluntary rather than a negotiated agreement.

I miss a bit of the part of actually is being doing for the sustainable part because when we are so concerned about digital sovereignty and actually creating more infrastructure for better data centers, better you know like just having more data power if we don’t think of the other side of the balance and how we are actually promoting that and for sure green energy like green data centers is that concept really exists is definitely part of it but even like environmental impacts because it’s not only about using you know like okay Brazil use a lot of energy from water sources which supposed to be cleaner for sure but still there’s a lot of environmental impacts that sometimes we create and this should be part of the discussion

speaker

Thiago Moraes

reason

This comment highlights the tension between digital development goals and environmental sustainability, bringing attention to often overlooked environmental impacts.

impact

It broadened the discussion to include more emphasis on environmental considerations in AI development, which the organizers acknowledged as an important point to address under the ‘planet’ aspect of the declaration.

So I don’t see the representation of somewhat of a general point of looking at, well, if I want to promote a certain point, I also have to incorporate others, as well as, especially if you look at AI and that most of the use cases are based on the data, obviously. And I mean, that is probably the area where we are lacking most in responsibility and in humanity, and whether this will also be part of it to look at the development point from a more holistic point of view and incorporating those SDGs, as well.

speaker

Claire

reason

This comment emphasizes the need for a holistic approach that considers potential conflicts between different SDGs and highlights data as a critical area for responsible development.

impact

It prompted acknowledgment from the organizers that they would need to consider a more overarching framework beyond the five P’s to address these interconnections and data issues.

And I think a human rights based approach to whatever you are doing, be it AI or the SDGs, actually, is very helpful in that sense, because if you make sure that you have a human rights based approach, you don’t only cover those very obvious elements such as discrimination, AI bias, all of that, if you indeed incorporate it in the design, development and deployment phases.

speaker

Trine (Danish government representative)

reason

This comment introduces the importance of incorporating a human rights-based approach into AI development for SDGs, providing a framework that can address multiple concerns.

impact

While not directly addressed by the organizers, this suggestion added a new perspective to the discussion on how to ensure responsible AI development across multiple dimensions.

Overall Assessment

These key comments shaped the discussion by clarifying the scope and goals of the Hamburg initiative, highlighting important tensions and considerations in AI for development (such as environmental impacts and potential conflicts between SDGs), and introducing frameworks like human rights that could guide responsible AI development. They pushed the organizers to think more holistically about the initiative and consider how to address complex, interconnected issues in the declaration and conference.

How to balance digital sovereignty and infrastructure development with environmental sustainability?

speaker

Thiago Moraes

explanation

Important to consider environmental impacts when building data centers and digital infrastructure for AI development, especially in developing countries

How to ensure access to AI infrastructure and hardware in developing countries, particularly the global south?

speaker

Audience member (unnamed)

explanation

Critical for inclusive AI development and addressing SDGs in underserved regions

How to develop AI governance structures in developing countries, particularly in the Middle East?

speaker

Audience member (unnamed)

explanation

Necessary for responsible AI implementation and addressing potential risks

How to incorporate a human rights-based approach in AI development for SDGs?

speaker

Trine (Danish government representative)

explanation

Ensures comprehensive coverage of issues like discrimination, bias, health, and environmental sustainability

How to showcase local initiatives and small-scale use cases of sustainable AI applications?

speaker

Thiago Moraes

explanation

Provides concrete examples and potential for scaling up successful projects

How to ensure representation of voices from the Global South and indigenous communities in AI and SDG discussions?

speaker

Robert Opp

explanation

Critical for understanding on-the-ground realities and ensuring inclusive technology development