Leveraging the UN system to advance global AI Governance efforts

UN experts discuss AI’s role in sustainable development and the need for inclusive governance

At a recent session focusing on the intersection of artificial intelligence (AI) and the United Nations Sustainable Development Goals (SDGs), experts from various UN agencies discussed the potential of AI in advancing global development and the role of the UN in fostering AI governance and international collaboration.

Reinhard Scholl opened the discussion by setting the twin objectives of exploring AI’s contributions to the SDGs and the UN’s support for AI governance. Doreen Bogdan-Martin from the International Telecommunication Union (ITU) emphasised the importance of a unified UN approach to support countries in leveraging AI for the SDGs. She highlighted the need for inclusive and equitable development of AI, which requires bringing together member states, academia, civil society, and the private sector. Bogdan-Martin also mentioned the launch of the UN AI Activities Report, showcasing the significant increase in AI initiatives within the UN system.

Tshilidzi Marwala from the United Nations University addressed the digital skills gap, particularly in the Global South. He identified four key areas that need attention: access to data, expertise, computing power, and good applications. Marwala called for a platform where the Global South and Global North can co-create effective AI solutions.

Gilbert Houngbo from the International Labour Organization (ILO) discussed the impact of AI on jobs, acknowledging both job losses and creation due to AI advancements. He highlighted the importance of skilling, re-skilling, reconversion, upskilling, and lifelong learning to minimise job displacement. Houngbo also raised concerns about the lack of data on the informal economy in low-income countries and the need for the ILO to address the peculiar challenges it presents.

Daren Tang from the World Intellectual Property Organization (WIPO) discussed how the UN can balance intellectual property rights with promoting innovation and equitable access in AI governance. He advocated for an ecosystem approach that ensures AI benefits are distributed globally and mentioned that the IP system should remain human-centred, not recognising AI-driven creations without human involvement.

Dongyu Qu from the Food and Agriculture Organization (FAO) discussed the need for increased food production and the role of AI in achieving this as the global population grows. Qu stressed the need to produce more with less: more quantity, high quality, more food diversity, with less input, less negative impact on the environment, and less social injustice. He called for more dialogue between biologists and IT professionals to bridge the gap between biological intelligence and AI.

In the final round of questions, the panellists offered recommendations for the UN’s role in AI governance. Daren Tang emphasised the need for inclusivity, suggesting the UN should serve as a platform for global dialogue. Tshilidzi Marwala recommended creating a co-creation platform for diverse global participation. Dongyu Qu advocated for a positive, cooperative, and holistic problem-solving approach. Gilbert Houngbo highlighted the importance of UN leadership in coordination and ensuring that AI does not exacerbate inequalities. Doreen Bogdan-Martin concluded by calling for leveraging collective assets, better coordination, inclusivity, and adherence to UN values at the heart of AI efforts.

Reinhard Scholl:
Okay, thanks very much, ladies and gentlemen. We have two goals in this round, in this session. One is, what can AI do to help achieve the United Nations Sustainable Development Goals? And the second part is, how can the United Nations system support AI governance efforts and international collaboration? So we have five number ones here. That’s pretty cool. Thank you very much. Two of you, I know, have a pretty sharp deadline at 1600 hours. So there is nothing what my colleagues cannot do. They could probably also get you a helicopter, but maybe let’s do the following. We do the first round of questions as planned, and then the second round of questions, I’ll take the two colleagues who have to run first. And don’t feel shy about just leaving, because you have other important meetings coming up. Okay. First one is, I’ll turn to the International Telecommunication Union, to Doreen Bogdan-Martin. So what can the United Nations systems do to ensure that AI supports the Sustainable Development Goals?

Doreen Bogdan Martin:
Thank you. Thank you, Reinhard. And good afternoon again, ladies and gentlemen. You know, I think, to answer your question, what can we do, I think the proof is actually right here on this amazing panel, and I’m delighted to be joined by my fellow UN colleagues, because what countries need, what countries want to see, is actually the UN working as one on the ground, and we’ll hear this afternoon a couple of great examples. I think when we look to the 2030 Agenda, and we remind ourselves that we need to leave no one behind, that it’s fundamental that the UN system steps up and engages. The pillars of the 2030 Agenda are focused on people, focused on planet prosperity, peace and partnership, and of course, AI is at the core of all of those pillars. And you know, the UN has this ability to convene, convene its 193 member states, and of course, on top of that, we need to bring in academia, civil society, private sector, to ensure that we have inclusive discussions. I think the goal here, and we heard this also from the previous panel, is we want to have inclusive, equitable development of AI. And I don’t think it will happen automatically, we need to make it happen, and we need to make sure that all voices are brought to the table. I think we’re doing a lot already as a system, including, of course, the AI for Good platform, which brings together some, I think, 27,000 or 28,000 experts in 180-plus countries. We need to leverage this platform, which brings more than 40 UN agencies also together in this context, and we need to leverage that and move it forward. I wanted to mention that actually today, we launched our UN AI Activities Report, and it’s really exciting because we’re actually up to some 400 initiatives, projects that are ongoing within the system. Some of them are internally focused, like on data sets, data management, dashboards, but many of them are externally focused about delivering to our member states, and that’s where we’re all working together. And that report, actually, the number of projects have almost doubled since last year, which, again, demonstrate that there’s absolutely much happening. Of course, we work with WHO, with WIPO on health. We launched that effort last year. We’re working closely with ILO on everything labor, and, of course, we have our new focus group on agriculture and looking at artificial intelligence and agriculture. Something also that’s new, and hopefully Professor Shiladze will talk more about it, is our new initiative with UNU, where we’ll be advancing a flagship report that will leverage and take advantage of all the research and knowledge and case studies that we have gathered at this forum and put it together in a way that’s easily digestible. It could be a good tool, and that will be, I believe, an annual effort. I did also want to mention, but I won’t get into the details, because we’ll hear more from the co-chairs later today, we also have this very important interagency mechanism, a working group on artificial intelligence that the ITU has the pleasure of leading with UNESCO. So, really, a lot’s happening. We have to leverage AI so that we ensure that we can achieve the SDGs, that we leave no one behind. It’s not going to be one agency that can do it on its own. It’s the whole of the system that has to come together to deliver and leave no one behind. Thank you.

Tshilidzi Marwala:
Thanks very much, Doreen. Turning to the United Nations University, Mr. Chiditsi Marwada, so how can the United Nations work with governments, with schools, with industry to close the digital skills gap and make sure that everyone benefits from an economy that seems to be more and more dominated by AI? Thank you very much, and congratulations, Doreen, for really leading us. Now, how do we work with, what are we trying to do? I think on AI, there are four things that we need to pay attention to. I think the first one is data, access to data, especially in schools in the Global South. I spent much of my career in the Global South. The second one is access to expertise that will be able to teach some of these algorithms. I think that should not be taken for granted. Then the third aspect is access to computing. It’s very expensive to buy cloud computing, and I’m speaking from experience now. If you go to Amazon Web Services, you will realize that it becomes very expensive very, very quickly, and many people in the Global South do not have access to high-performance computing. Then the fourth aspect is access to good applications. And of course, when you are going to apply this technology, you need infrastructure for data collection, for data storage, infrastructure for use. That is actually quite an important aspect. As the United Nations, I think there are several issues that are at play that are going to address some of these issues. I think the Global Digital Compact is going to deal with these issues. I actually wrote on Fortune magazine that maybe the next decade should be declared the International Decade of Data, because without data, none of these things are happening. Now, what are some of the platforms that we need to put into place so that people in the Global South have access to some of these things that I’ve talked about? I know of an initiative in Africa called Deep Learning Indawa that has been teaching people how to do these things. And then, of course, the High-Level Advisory Board on AI will come with recommendations. But I think we have lots of ideas. We don’t have a crisis of ideas. I think what is important is how do we make these ideas work? And I think the way we do that, we need to change human behavior, you know, behavior of users, behavior of teachers. We need to create incentives so that these technologies are taken into account. We need to create institutional structures, and we need to have proper policy framework. So the idea, Doreen, you have worked quite a great deal on this, building capacity on policymaking on these issues becomes extremely important, and we should expand it. Thank you.

Reinhard Scholl:
Thanks very much. Moving to WIPO, the World Intellectual Property Organization, there in town. So how can the UN balance intellectual property rights on one side, and then also promoting innovation on the other side, and equitable access in AI governance, how do we do that?

Daren Tang:
Thank you, Reinhard, and thank you, Doreen, for leading us on this important conversation. Very happy to meet my fellow UN colleagues, working horizontally to support the member states. Look, intellectual property is not an end in itself, and we have to stop looking at IP as an end in itself. We have to look at IP as a means to an end, and what’s the end? Creating jobs, supporting entrepreneurs, supporting people who have got the hunger to pick up skills and need those skills to bring their ideas to the market, and ultimately as a catalyst for growth and development. So I think across the UN family, we are working towards this common vision in the global digital compact and the other documents that we are going to discuss at the Summit of the Future in 100 days’ time. And I think IP really can be a very powerful enabler of all this, and I can understand why there’s a conversation around access, and let me give you some statistics. In 2022, 3.4 million patents were filed around the world. Of the 3.4 million patents, 60,000 were connected to AI, not just gen AI but all different types of AI. And so the question then becomes, how do we help to promote that this innovative ideas that’s being developed by people, primarily in the more mature and developed economies, can be a source of inspiration, a source and a way to support those in the global south as well. And I think what that means is that you need to create, as Zilli CSA, an ecosystem where that technology can be absorbed meaningfully. If you take a glass of water and you pour it on soil that’s very dry, it doesn’t absorb, and the water just flows away. But if you pour it into a soil that’s already been prepared, then the technology can be deployed in a meaningful way because people have been trained, the computing power and access to computing is there, the opportunities are translated into actual things that happen on the ground. So this really needs an ecosystem approach and that’s why it has to be horizontal. It means that WIPO needs to work with ITU, with UNU, with ILO, with FAO to make sure that we bring all these elements together. And so I think that’s really critical. And the other thing about the IP system that not many people know about is that it is a human-centred system. The IP system is a system that puts the human creator and innovator at the centre of it. So in other words, if it’s entirely driven by a machine, the IP system will not recognise that. And I think that’s a good thing because we need to make sure that AI and technology enhances and supplements and makes us more human, if that’s possible, and I think that’s the vision that we should have, rather than undermine and undercut and make the human being irrelevant. So I think the IP system is in a good position. As we transform it to make it more inclusive, to emphasise its role as a catalyst, to really play a very important part of this global conversation on how we can really make technology and digital technology an enabler, not just for the countries that are able to do it now, but for the world in general and for the global South. So thank you very much for that. I look forward to more discussions. Thanks very much.

Reinhard Scholl:
And moving to the International Labour Organization, ILO, Mr Gilbert Hombou. Everyone is worried about jobs, the impact of AI on jobs. Can you help us? Easy question.

Gilbert Houngbo:
I like the question. First of all, let me thank Doreen. I want to join what my colleague was saying at the beginning, that this initiative of getting the UN, colleagues from the UN together, and this coordination and harmonisation is quite critical, because there are so many dimensions from the governance or the safety security dimension that the previous panel was discussing. With our sectorial perspective, I believe it’s key. Clearly, one of the key concerns is what are the impacts of AI on jobs. So clearly, even 10 years ago, ILO started doing a lot of analysis and research on the impact of technology advancement on the jobs. And clearly today, we published some data almost eight months, nine months ago, which we keep updating. So it’s very clear for us that the short-term concern are the jobs that have already been lost. So it’s not about the risk of losing jobs. Jobs have already been lost, and we will continue having millions of jobs that will be lost. But we also know, at the same time, that we’re going to have millions of jobs that will also be created. So it’s very important to keep the two sides of the coin together. is one. And when you look at that, I think it’s important to put things into perspective. It’s not different than the different industrial revolutions that the world has gone through. And recently, maybe, you know, when you think something used to be called typewriter, which we don’t, nobody really, our kids will not even understand what it means today on that. So our point is that between the jobs that are going to be lost and the jobs that are going to be created, a lot are really going to be transformed or enhanced by the technology, enhanced by the AI, hence the famous expression of augmentation. So I usually would tell people that, of course, mathematically, you can think that if the new jobs being created are more than the jobs that have been lost, we’re not bad. Except that you don’t want to be in the category of the job to be lost. None of us want to be in that category. So the key thing is what? The key thing is scaling, re-scaling, reconversion, upscaling. That is in the short run that ILO is really advocating. And I wish one day we could also have our colleague from UNESCO here. The other dimension that we are noticing is that we all have to step up the lifelong learning. The lifelong learning is going to become key for us to minimize the impact on the job, which is one. On the medium to long term, we do have concern because that’s where I really want to join you with the data issue. And I can say from a perspective, a lot of the economy from the south, a lot of low-income countries, 70%, 80%, 60% of the economy is informal. And there is absolutely very little data for us to be able to assess the impact of AI on the informal jobs, informal economy, those small businesses that are thriving on a daily basis. So our research department is working right now with colleagues with the World Bank, trying to really see how do we again address that peculiar challenge that we have. A lot of data and a lot of statistics that we are talking about is coming from the global north, if I can use that term, on that. So it’s important for us to really also see how, therefore, we do have the data from the working population, as I said. The other point that I think is going to be crucial for us, you know, maybe the day will come where AI will help us increase all the productivity that the number of jobs required may be way below the size of our labor market. In other words, if today the delta is positive in terms of new jobs created versus jobs being lost, right now it seems to be rather positive. But maybe in the long run, that day could come where, so one of the struggles we are going through in ILO is what job would mean and what would that mean for us, the value itself. And one of our key principles in ILO is that labor is not and should not be a commodity. But the time will come where maybe we’ll have more time than the work that we need to do, on that, when you look at it positively. Last point is the, again, I’m back to short term now, next year 2025, in ILO we’re going to start what we call the first of a double discussion on the impact on the platform economy, on the platform economy on the job market, and what to do, again, to encourage the positive dimension of the platform economy, the job creation that comes with it, the wealth creation that comes with it, and associate that with the necessity to also have a minimum respect for labor rights, for human rights, as well as the protection, the social protection for workers.

Reinhard Scholl:
Okay, thanks very much. I’m turning to Mr. Zhu Dongyu from the FAO, the Food and Agriculture Organization of the United Nations. We all need to eat. Can we use AI to eat better, even better, or better, that we all have enough food, that the food is distributed equally? What do you tell us?

Dongyu Qu:
Thank you. Thank you, ITU, and again, this platform for the UN system to debate a little bit. First of all, I would make some data or evidence based. We need more food because 2050 we will have a 10 billion population, and more and more people stay in the city, urbanization, 70%. So that means less farmers, so we need to produce more with less. More quantity, high quality, more food diversity, and with less input, less negative impact on the environment, less social injustice. That’s the agricultural food system and the rural demand that we share the common vision. And then how? The AI is a new cutting edge or tools. We need all the necessary possible potential tools in the toolbox. IT in general, BT, biotechnology, and others, new materials, new equipment, you name it. Yesterday, I had a very long discussion with the president of EFPL here, because I’m a biologist. You are AI. What means AI? Artificial intelligence. We have to build bridges between AI and the real AI from the human being, from here. It’s most complicated animals, intelligence animals on this planet is human beings and biology. So we need more dialogue first. Second, food security. We need food availability first. It means we have to improve the productivity with innovation, with enabling policy, and also the investment. And then food accessibility. We need the logistic, and we need also all the value chain management. And then the food affordability. So we need also reduce the food cost and reduce the food loss and waste. So that’s all the general ideas, but AI can play the role. How to improve efficiency, effective, and also productivity. So that’s why I’m so interested during the past 40 years, even 30 years when I was a scientist, to try to bridge biologists with IT people. You face the computer. Why are you cheating computer? Computer will cheat you. Cheating. So a lot of people, they didn’t have enough data, or only limited data, and they got a conclusion. And then you get other sectors for application and so on. So he was an expert. He’s an expert on the IT. I was an expert on BT, genetic engineering. We need a more close cooperation. And for that, FAO, I established an open platform for all the scientists to debate, science innovation forum, and every year, October, middle of October. Second, third, and also we are first to launch the AI ethnic call, Roman call, together with IBM, Microsoft, and the Holy See. So we need the ethnic because so many years, no matter developed, developed countries have a digital gap or divided. We need to fill in the gap with the infrastructure, improvement, broadband, accessibility, you name it. And also we need the training farmers. So that’s why we need to work together with education, UNESCO, and also with WIPO here. And also, fortunately, he was a former president of the IFAD. He understood the rural affairs. So I think for that, if we’re willing to cooperate with you, SDG1, SDG2 is most important. You need food. You need bad food. Without food, we are not able to work. You can’t eat AI. You have to eat the food. Thank you.

Reinhard Scholl:
Okay. Okay. Thanks very much. All right. Well, the second question, which is what can the UN do with respect to AI governance and international collaboration? One recommendation from each of you, one minute max. And we start with Teran Tang because you have to run off and don’t feel shy just to leave. So what is your one recommendation to the UN system with respect to AI governance?

Daren Tang:
I think the most important thing is that we need to be the platform where we are big tent and we’re inclusive. I’ve seen since GEN-AI came out last year, a lot of meetings and conferences, but mainly in developed countries, nothing wrong with that, of course. But my worry is that when the conversations are within a small group of countries and you leave out the majority of where the population is, you are finding solutions that people get alienated from because they’re not involved in shaping them. And we know that trust is a major issue, not just in AI, but in the world right now. We are facing a multilateral system where consensus is extremely hard to find, where negotiations don’t lead anywhere. So if we were to set rules with just a few countries at the negotiating table, and most of the subjects of that will be outside of that negotiating table, I think we’re going to have a situation that is not good for anyone. And one of the things that I want to just leave with all of you is that different data points show us that, in fact, the hunger for AI is stronger in the global South. There’s a Stanford AI survey that showed that attitudes towards AI was most positive in Malaysia, and I think Nigeria, and a few of these countries, and most negative in the developed countries. And WIPO did a survey last year called WIPO Pulse on attitudes towards IP, and attitudes towards IP was the most positive in Asia and Africa. So actually the hunger is there, so it’s just that we need to bring them in, and I think the UN can be the big umbrella, the big tent to have these inclusive conversations.

Reinhard Scholl:
Thanks very much. Mr Mawala, one recommendation, one minute.

Tshilidzi Marwala:
Well, I think one recommendation is that we need to create a platform where people from the global South and people from the global North can come together and co-create, because it is quite clear that if we leave it to one section of society, we are not going to have an effective AI that works for everyone.

Reinhard Scholl:
Good, thank you. Mr Chu, one recommendation, one minute.

Dongyu Qu:
Only one? It’s okay. One minute. Be positive. Be cooperative. Solve the problem holistically.

Reinhard Scholl:
Thank you. Okay, thanks. Mr Hongbo, one recommendation.

Gilbert Houngbo:
I think, first of all, the UN has to take leadership, and I want to count on Doreen. First of all, this coordination within the UN is crucial, and secondly, between the UN and the rest of the multilaterals, as well as the non-multilateral key stakeholders, namely the private sector, the academia, the bilateral government, the public sector, which is going to be key. Thirdly, I think the UN system will be best placed to ensure that nobody is left behind, that this is not another source for creating more inequalities.

Reinhard Scholl:
Okay, thanks very much. Doreen, you have the last word.

Doreen Bogdan Martin:
Can I respond to Professor Shimizu, and then I’ll give you my one recommendation. So I fully agree with your proposal, and I think in terms of bringing the global South and the global North together, this is where it happens. I think that co-creation is happening here today at Governance Day and over the next two days, so I think we have to build on that platform. And I guess my recommendation is, let’s build on our respective and collective assets. Let’s better coordinate. As Thomas Schneider said, let’s be more coherent in our approaches. Let’s make sure we’re inclusive, and let’s make sure that we don’t further exacerbate the divides. I think we always have to remember, if we don’t bridge the digital divide, we’re always going to have an AI divide. So let’s keep that at the core, and let’s put the UN values in the center, at the heart of all of our efforts. Thank you.

Reinhard Scholl:
Okay, thank you very much.