WSIS Action Line C10: The Future of the Ethical Dimensions of the Information Society

UNESCO panel explores ethical AI, disinformation, and the need for global governance

During a UNESCO session focused on the interplay between artificial intelligence (AI), disinformation, misinformation, and ethical AI, a panel of experts convened to explore these pressing issues. Dr. Mariagrazia Squicciarini of UNESCO moderated the discussion, which included insights from Ashutosh Chadha of Microsoft, Amanda Leal of the Future Society, Fadi Daou of Globethics, and Ricardo Baptista Leite of HealthAI.

Ashutosh Chadha emphasized the widespread accessibility of AI and the critical need for ethical frameworks to guide its use. He stressed the importance of keeping humans at the center of AI development and the potential of AI to address challenges like misinformation and cybersecurity. Chadha advocated for risk-based, outcome-focused, and collaborative approaches to AI governance.

Amanda Leal discussed operationalizing responsible AI frameworks and the importance of AI governance in shaping regulatory frameworks and policies. She highlighted the need for accountability, transparency, and stakeholder engagement, emphasizing the role of oversight in addressing information asymmetry and fostering trust in AI technologies.

Fadi Daou addressed the ethical challenges and opportunities AI presents, particularly in relation to bias and socioeconomic divides. He argued for the importance of partnerships across sectors to ethically navigate AI’s potential and called for AI literacy and digital literacy to empower people against misinformation.

Ricardo Baptista Leite focused on AI’s transformative potential in healthcare, cautioning against retrofitting AI into the existing disease model and advocating for a healthcare system that leverages AI for prevention and health promotion. He also highlighted the role of governance in building trust and the importance of HealthAI’s work in supporting governments and connecting regulators globally.

The panelists collectively agreed on the importance of ethical AI that respects human rights, dignity, and freedoms. They discussed UNESCO’s role in developing tools and recommendations for the ethical implementation of AI, including readiness assessment methodologies and ethical impact assessments. The conversation also touched on the need for diversity in AI programming and the potential of AI to impact the environment.

Audience questions raised concerns about the ethical responsibilities of companies like Microsoft in releasing AI technologies that could be misused, as well as the importance of considering AI technologies in the plural due to their diverse applications and impacts. One audience member questioned the effectiveness of international organizations like UNESCO in defining and enforcing ethical standards.

The session concluded with a call for collaborative approaches involving governments, the private sector, civil society, and international organizations to address the ethical challenges posed by AI. The panelists emphasized the urgency of addressing disinformation and misinformation, particularly in the context of democratic processes and elections, and the need for public awareness, literacy, and capacity building to empower individuals and uphold the rule of law in the age of AI.

Noteworthy observations from the discussion included the recognition of AI as a present reality rather than a future concern, the need for a multi-stakeholder approach to steer AI ethically, and the potential of AI to serve as either a great equalizer or divider based on how it is governed and deployed. The panelists also acknowledged the challenges of translating policy thinking into actual policymaking and legal frameworks, as well as the importance of embedding inclusivity, fairness, transparency, and accountability in the design and deployment of AI technologies.

Dr. Mariagrazia Squicciarini:
Good morning, everybody. Welcome to this session that actually tries to discuss the issue that how AI links to disinformation, misinformation, and actually what it means to have an ethical AI addressing the problems. My name is Maria Grazia Squicciarini. I’m the chief of the executive office of the social human sciences sector at UNESCO. As you see, there is nothing short about my name or my title, so it sounds like that. And actually, we still miss one or two speakers, but we’ll start nevertheless, because this topic is too pressing and too important to delay the discussion. And I’m going to ask my panellists actually to introduce themselves very briefly. Your name, your title, and actually why you think the discussion is very important. Thank you.

Ashutosh Chadha:
Thanks. Ashutosh Jadhav. I’m the senior director for Microsoft here based in Geneva, and I’m part of the UN International Organization for AI. Why is this topic important? I think given the democratisation of AI, which is in the hands of each and every person over the last one year or so, it’s not about what AI can or should do, but it’s about how do we make sure that we leverage the potential of AI ethically and mitigate all the problems that may come up in future. So that’s where I think why this conversation becomes extremely important.

Dr. Mariagrazia Squicciarini:
Over to you.

Amanda Leal:
Okay. I’m going to do some timekeeping. Hello, everyone. I’m Amanda Leal. I work at the Future Society, a non-profit, and I’m here as an AI governance professor. civil society representative and also as a Brazilian, because this is how I, I think it’s important to mention, I think how I got into this field. It was back in 2018 when we were navigating the most turbulent elections we’ve ever had, and that’s how I was parachuted into legal and policy research. And at the Future Society, we work to advance governance mechanisms, regulation, and policies and standards for AI governance. And so I think this is a pressing issue and what I’m looking at at the moment is how to operationalize responsible AI frameworks and how to advance regulatory frameworks and inclusive governance. So it’s a pleasure to be here.

Fadi Daou:
And we can guess what you will do. Yeah. Thank you. I’m Fadi Daou. I’m the Executive Director of an international organization which has offices in Geneva called Globethics, having regional offices in the world. So we work at Globethics for ethical leadership, and one of four priority areas is digital and emerging technologies, and you can imagine why I’m interested in the question of AI. I do believe that, of course, there are ethical challenges to be taken into consideration when it comes to AI, but also we like to look at AI as an opportunity also to face some serious problems in our, I mean, even social and societal problems that AI can also help. We’ll talk more about them later, like bias or socio-economic environment.

Dr. Mariagrazia Squicciarini:
Thanks a lot. By the time our third speaker will arrive, fourth speaker will arrive when we introduce. Now, as you can see, I’m really happy to see this room packed for two reasons. It means it’s important. And actually the topic is not as big as the air, so perhaps we might want to leave a bit of the door open. But let’s go now into the details. As you mentioned, this is about the future we want. Now, I actually disagree with you. This is about the present we already are living in. Because very often, by the time we talk about AI, there is a… tendency to say two things that for me are twofold. The first one is a black box. Nothing can be done about it by the time it’s out. By the time it’s out, it goes its own way. Off you go, you can’t fix it, kind of, out of our hands. And the second thing is to have this concept about the ethics of AI as an exposed thought. Like, we first develop, deploy, use AI. Then afterwards, we decide, what can we do with it? How to be ethical? And I mean, ethical for UNESCO means something very, very concrete. It means respecting human rights, human dignities, and fundamental freedoms. And why UNESCO believes that’s fundamental as a problem is because this is a technology that knows no boundaries. We need to have definitions that actually work regardless of where the technology is developed, or where it is employed, or where it’s deployed. And therefore, if we start going by, oh, but my culture says this, oh, but my beliefs say that, of course, we start offering a way for those that want to develop the technology or use the technology in the wrong way, to act for each other, to leverage and develop it or use it in places where, let’s say, the protection of some of the values are different. What UNESCO did, and we were so proud to see our 193 member states agree on a text that, I don’t know how many of you have read it. I’m not going to ask, because it will look like at school, like, have you read it, the material? But actually, the recommendation of UNESCO is not only very wide in terms of the topic, it touches upon, there are 12 policy chapters that range from privacy to cybersecurity to skills and education, but also the debt. So there are provisions in the recommendation that are about not only do not harm, but there is the redressal mechanism that is, of course, things can go wrong. Any technology was developed with a goal and ended up in the wrong direction at some point in time. What is important is actually to be able to reach and stop from doing harm. There is also a principle that’s not very often we hear talking about, and this proportionality. I bet in this room, you all have experienced that you’re exposed to whatever apps are used where you have actually, they say there is AI, and then the question becomes, do we really need AI for this? You know, a normal algorithm without any learning would do a perfect job. So the overshooting problem, like now everything needs to be AI guided. And the other thing that is, let’s say, perhaps the third tool’s legend that we hear, is a conversation that tends to be actually focusing on the technology itself, which is actually something that then loses half of the discussion, because not all of us we are experts in AI, of course, we have our own limits, we are specialized in different things. But what we believe at UNESCO should be the approach, is a conversation about the societal aspects of the technology, what we want this technology to do for us and what we don’t want it to do. I think we have our four families in the meantime. Yes. Please, if you could say that very briefly.

Ricardo Baptista Leite:
Oh, sorry for coming in late. I was lost in the building. Welcome to my class. I was as well. And so I’m the CEO of HealthAI, the global agency for responsible AI and health. Our office is across the street. If you need a break, you can come visit us. I’m a medical doctor treating infectious diseases and a former member of parliament from Portugal and also the founder of the United Parliamentarians Network for Global Health, which is present in 112 countries. It’s a pleasure to be with you. Thank you so much.

Dr. Mariagrazia Squicciarini:
Thank you for joining. So actually we are at the point of discussion where I would actually ask you, now let’s go back to the topic of today, disinformation, misinformation, and what, in my view, ethical AI for the purpose. I would invite our panelists to briefly expose, I will give you two, three minutes and then we’ll take another round. And please, if you have questions, just let me know and we will try to squeeze them in in a timely fashion. And. Actually, it would be interesting to not only look at the problems that AI might add on to the already big issues we see with disinformation and misinformation, but also how AI can actually help finding possible solutions to that and what it can contribute. So perhaps we start again from you, Ashutosh.

Ashutosh Chadha:
Okay. No, I think, and I’ll actually twist your question a little bit, or possibly, I think you’re very right saying it’s not about the future, it’s about the present, right? And one of the things I think all of us should be pleasantly very happy about is that if we see what’s happened over the last one year, one and a half years, ever since generational AI exploded onto the firmament or on the planet, so to say, we’ve had a lot of conversations around the right way to use AI, the guardrails. So whether it’s conversations which are happening at G7, whether it’s the UNESCO framework, whether it’s G20, whether it, you know, everywhere we’re having these conversations. And I think that possibly is the most redeeming feature of the discussion today, because it is important to keep the humans at the center. And I would possibly frame our positioning into two ways when we start looking at what should we be doing? It needs to be risk-based, it needs to be outcome focused, it needs to be what I would say collaborative in nature, whatever framework we are building. And most importantly, it needs to have humans at the center, both humans with the ability to control the technology and built with humans from a design perspective, which you mentioned. And I think when I look at that perspective and say, what can AI do to help us address all of this? I think there is a technological solution which are available. I mean, you talked about misinformation, disinformation, there’s a lot that AI can do to actually address that. We talk about cyber security issues. There are national security issues and there are also cyber security issues which are related to phishing mails and people getting spoofed and having to lose money. It’s all over there, right? AI can help with that. So there are technological solutions. But I think so the greater good, and this is possibly where I’ll sort of stop for the others. The greater good would be when we all come together, right? Build those tools and implementation resources to take all this conversation that is happening across the globe, right? And help countries and organizations implement. So for example, the tools that the RM that UNESCO has come up with. And if we can make sure that they really get implemented. And that in my opinion is the biggest challenge. The second biggest challenge is threading the needle between all these conversations which are happening. And the third biggest challenge is to make sure that the private sector, civil society, governments are actually forthright in saying what are the potential downfalls of this technology and how should we control it? The reason why I’m saying that’s very important is because if you look at cyber security, one of the biggest problems in cyber security is that people don’t talk about it when they have been breached. Because they feel that you’re going to go after them. You cannot hide behind that cloak of secrecy, not in this technology.

Dr. Mariagrazia Squicciarini:
Now, if I were the genius of the lamp, I would say you have no desires left. In the sense that, yes, UNESCO is working very actively about the implementation of the recommendation. And for those that don’t know, in the recommendation it tells there are two tools that actually countries ask us to develop. The readiness assessment methodology that you refer to, that is about helping countries take a snapshot of where they are in terms of the infrastructure, the legislation, the components, the actors, the leading actors. And this is not only government, it’s really about all the relevant stakeholders, companies, et cetera, that actually are needed for AI to be developed, deployed, and used ethically. And you also said multi-stakeholders, indeed, and this is why we have you here today. And these, we have not only developed the recommendations together with relevant stakeholders, but now in the implementation, we are actually working with the relevant stakeholders because I think what has been bad in the past, which makes also create tensions that are really not needed at this point, is the fact that very often civil society organization comes to complain about something that doesn’t work. But since they don’t get involved in the design of the solutions, of course, they might just complain afterwards, but that doesn’t help solving the problem. So the approach of UNESCO is an excellent approach whereby we actually try to involve all the relevant stakeholders. I mean, true to be told, Microsoft is actually co-leading what we call the business council. I mean, let’s be truthful. There cannot be any steering of the technology without the involvement of the business sector. It just wouldn’t work. It’s just the technology that the private sector is mainly developing. So what we discussed, and we were very pleased to see the engagement of Microsoft among others, because this business council sees the involvement of many leading companies, just LG is another one or Mastercard is another one. So we have a few that are very big, and the co-chair is Telefonica, is really to have them, the companies themselves, help us finding a way to implement the recommendation in companies. Because, of course, what countries need to do in terms of legislation, institutions, et cetera, is very different from the kind of checks, the kind of safety design that companies need to do, because they have hundreds of products embedding AI, developed and employed at the same time. So, you know, the typical check afterwards won’t work, especially by the time that these products go into the market. And we have another tool inside the recommendation, which is the Ethical impact assessment, that is exactly the kind of thing that we are now playing and molding in order to be used in companies, that actually aims at one specific thing, that is, here the interest of nobody, I suppose, is to stop the development of the technology, because AI can do many fantastic things for us. But, of course, we want to do the things we want to do and not to harm us. And so the idea of the ethical impact assessment is actually to check what comes out. Now, we can discuss whether risk-based or not, that’s a different discussion, and make sure that nobody’s harmed. And by the time there is harm, it is redressed. And there is also a responsibility, because I think none of us is happy by the time they say, oh, the computer doesn’t allow to put your name. I mean, I’ve experienced that several times with my long name and surname. But this is just a silly thing. But imagine what it can be, and I’m going down then to Ricardo, by the time this is applied to health, for the system says, oh, you’re not entitled to, or no, you’re not, you know, you know, you’re not among the sample that we will deal with. And of course, then we might be talking about surviving or not, being treated or not. And the level of, you know, the disinformation, because of the sample that has been used to decide leveraging AI, it’s not really representative of, for instance, the way Maria Grazia is. Ricardo, over to you.

Ricardo Baptista Leite:
Thank you. I had sent three slides to Rosanna, I don’t know if you’ve read them.

Dr. Mariagrazia Squicciarini:
I’m feeding them back to the system. Yeah, I mean, we are having an issue, so…

Ricardo Baptista Leite:
Okay, I’ll go, I’ll go for it. Yeah, good. Okay. So, well, thank you so much again. It’s okay, but I feel protected. And in Portugal, we say angels don’t have backs. So I’m not calling myself an angel. But it’s a real pleasure to be here, acknowledging the work UNESCO is doing in this space, which I think is incredibly important. And at Health AI, where we put ethics and equity at the core of our mission, I believe that we… experience in this effort early on to be able to address the challenges related to artificial intelligence as a transformational power in society. So at Health AI, we are the global agency for responsible AI and health. We are a nonprofit foundation. We were created with the institutional support from WHO, who sits on our board back in 2019. And we are basically an implementing partner for multilateral organizations like the WHO, and not only, but also for governments helping build regulatory capacity. I’ll get to that in a second. Just a step back, because I think we’re talking about the human factor here and the systems factors here at play. So from health care, we have an underlying problem before technology, which is we have a broken disease model. Basically, we call them health systems, but around the world, what we have are disease systems that basically react to disease. This has led over the last 100 years to a rise of costs to unsustainable levels, along with the rise of the burden of disease. So it’s a vicious cycle that if we do not break, we will end up with a two-tiered system. People who have resources have access to care. People who do not have resources do not have access to care. And so we’ve been calling for a change. WHO and so forth has been calling prevention, health promotion, putting quality of life at the core. But nobody has done anything. And I would say that partially because we didn’t have the technology to do so. AI represents the biggest opportunity for transformation in health that we have ever seen in our lifetime. And that being said, what we’re seeing by most technology developers, a lot of the startups out there are trying to look at AI and retrofit AI into the current model. And this, I think, is a lost opportunity. So the opportunity we have as humanity is to break down this preconception of a model based on hospital. and so forth, with the technology that we now are building, what can be the healthcare of the future? And that’s the exciting part, especially when we talk about universal health coverage. What we’re seeing is rich countries exporting the broken disease model to poor countries. And this is going to be unsustainable. Seeing huge hospitals being built in low and middle income countries where there are no healthcare workers. This is wrong. It’s a waste of money and most importantly, it’s a waste of an opportunity to put technology at the service of people that actually could get access to better care. And this touches upon the issue of inclusivity. And so, one of the big ethical discussions is, is it better to provide care through a community worker using an AI driven technology, but not having a doctor or a nurse or a pharmacist, or is it better not to provide any care at all? These are the ethical considerations we need to respond if we want to be serious about universal health coverage, especially within low and middle income countries, but also in high income countries in the most marginalized parts of the populations. I worked my first life as a physician in HIV and AIDS. I can tell you that in rich countries, we have parts of societies that do not have access to care. So universal health coverage is truly a global phenomenon. The second point I’d like to make, when I stepped in as CEO a year ago, we conducted a survey to legislators to technology companies, to academics, and the response was unanimous. We had the most common word we heard was the word fear related to AI and health. And the fear was stemmed from the lack of understanding of the technology, fear related to not knowing how to address it, fear of the potential harm that of course is hyped a lot by the media. And so we decided as an organization, doing a new strategy to put governance of AI as core in our mission. So to address governance, because without governance, we believe there are two problems. First of all, there’s. greater risks for citizens and systems, the use of technology irresponsibly. And the second risk is on the other side of the spectrum, which is in the absence of governance, people do not trust the technology and therefore do not adopt. Without adopting, we do not accelerate the adoption of technology. And that leads to not maximizing the benefits that we can take of the technology, which is our ultimate outcome, our ultimate goal, which is to ensure better health outcomes along the way. So to close, because I knew, I felt you were coming. Yeah, I saw, yeah.

Dr. Mariagrazia Squicciarini:
That’s also because it’s very hot. I don’t know what you think, but we have cookies.

Ricardo Baptista Leite:
Well, this goes to show that we need a bigger room next time. It’s a lot of interest. So at HealthyEye, what we’re doing is we do not do standards nor do validation of tools. What we do is we support governments in building the regulatory capacity. We certify the regulators and connecting the regulators together globally. So we have an early warning system. So if there’s an AI tool that has the potential to go rogue, everybody gets a red flag and we work together. And also with WHO putting together a repository of all of the validated AI solutions out there. And we can get into details after. Just to say we are mainly funded by governments, of course, but also philanthropies and partners. But there’s a book called, and I’ll end with this, called Power and Progress, that looks at technology over the last 1,000 years. And every time, consistently, technologists are saying technology is going to save the world. And most of the times that did not happen. And the reason was, and seeing a few examples where humanity was successful, was the cases in which inclusivity, fairness, transparency, and accountability were embedded in the way we deploy the technology and design the technology from the start. And I think that artificial intelligence, where we are now, represents that starting moment that together, and that’s why this is so important to work you’re doing, together we can make sure that we make this a success, not just for the few, but for humanity in general, and for future generations to come. Thank you.

Dr. Mariagrazia Squicciarini:
Thanks a lot, Ricardo. Actually, yes, at UNESCO, we do feel the pressure in the sense that our membership of the representatives were, let’s say, very worried, already in 2020, 2019 about the problem, when nobody was talking about this. So we have this beautiful group of experts, which is called Commerce, and then we have another one, which is IBC, these are all acronyms, but they are experts in ethics of new science and technologies, because this is core to the mandate of UNESCO. I mean, many of you, I’m sure, by the time I say UNESCO, you see a label on a monument, and this is UNESCO for you, or, you know, big programs for education. Well, there is an S in UNESCO, and that is for science. And science means that we have two actually big departments, one is, and we call it sector, one is for natural sciences. So for instance, UNESCO is the one coordinating the tsunami alarms. The biosphere, I don’t think many of you know this. And then there is the social and human sciences, that is the place I belong to, that actually deals, among other beautiful things, like fighting against racism and discrimination, fostering inclusion, the fair transition, on the ethics of new science and technologies. So this is what we do. Otherwise, UNESCO would be an eco, un eco, so it wouldn’t work the same way. Now, over to you, Amanda, on the same problem. Actually, none of you has mentioned one issue. I will wait for you to stop, and then I will ask the question. Go ahead.

Amanda Leal:
Okay, so thanks so much, Ricardo, for the help, because I wanted to tweak the question and make it about how AI governance can tackle difficult challenges, because I’m sure AI has potential. I’ve seen the technical side of AI. I worked at a research institute in Catamula. I worked with sociotechnical research as well, and seen how AI could be used, for example, in fairness, and how many principles that we discuss in AI governance can be translated in mathematical notions and can be tackled from a technical perspective. What I wanted to propose, coming from a place of wanting to propose a positive agenda, and I think something that I see a lot in responsible AI frameworks, is accountability. And I believe accountability is the best way to future-proof an ethics-driven AI governance. UNESCO recommendation, it talks about explainability, accountability as auditability and traceability. So I wanted to break down what I mean by that because the problem with accountability is that it’s also a bit of a fuzzy concept because it is more contextual, right? So I mean, three things, basically. Transparency, which we’ve been advancing a lot in terms of tackling it from a technical perspective as well. But in the sense of accountability, I think it’s important to understand that developers and deployers need to provide information of how a system functions, what data was used to train it, design choices. So it’s really from an upstream level. And what I really like about UNESCO’s frameworks, for example, the ethical impact assessment, is that it has a lifecycle approach. And so we do need to ask those questions early on. We wanna understand what data is used. We wanna understand also other aspects of the supply chain, like what kind of human labor was used to train the data. So transparency is really key and understanding if there’s redress plans, bias mitigation. But accountability doesn’t happen in a vacuum, right? It is relational. So oversight is really important. And I’m thinking here that AI developers and deployers must respond to increase about how a system works, its outputs, its outcomes. And they have to bear the burden of proving they have a high enough standard of transparency, of ethics, fairness, because there is an information asymmetry there. We don’t know how these systems are developed, right? And, but for that, I think a big challenge for AI governance is we need an effective channel for oversight. So how does this work, right? I think this is a governance challenge that should be welcomed by the industry, by civil society and governments. How do we think of this effective channels for oversight? because I’m not only talking about regulatory oversight. We do have the EUAI Act as an example, bringing in the EUAI office, and we’re gonna see how enforcement works and how they do oversight. But I’m also worried about stakeholder engagement, which is my third point, because we wanna bring in a broad range of stakeholders. And another aspect of AI technologies, and I’m thinking here, especially generative AI, general purpose AI, and powered by foundation models, is that they’re applied downstream across jurisdictions. As you mentioned, we need solutions that are a bit jurisdiction agnostic that will work. And for that, we can’t really just count on national regulatory bodies. We need to have some sort of, and so it’s important to include a broad range of stakeholders, and we can talk more about what that could look like in practice, and this is important for enforcement, so we make sure that we address that information symmetry. So I’ll stop there, because I’m curious to hear other panelists.

Dr. Mariagrazia Squicciarini:
Thanks a lot, Amanda. Before I go to Fadi, let me challenge one thing she said, and explain why UNESCO used the words ethical, not only responsible. Responsible is good. We all need to be responsible if we want to go through this world without doing harm. However, at UNESCO, we believe it’s not sufficient, and I make a parallel to what it might entail, for instance, to produce a t-shirt, right? Responsible means that you abide by the law that exists in each and every country. So what happens next, and we have had instances in the past, especially, now they are less frequent, perhaps, or less detected, is that then producers go to the country that has the most flexible legislation. You’re still responsible. If you’re in somewhere else, you’re, you know, abiding by what are the contamination limits, what are actually the labor laws, whereby I can employ or not a guy aged, or a guy or girl aged 14, and still I’m responsible. Now, come back to this place, or Europe, or any other place where, you know, we really try, like, every country strive to have the best regulation. and the best protection for its citizens, where the legislation and the protection is very high. Well, you wouldn’t hire a person aged 10 to sew your T-shirt. Whereas many of us, unfortunately, without knowing, we buy T-shirts that are done in countries where the legislation says, you know, better to have someone that eats every day because he or she works, still if it’s 10. Now, this is why UNESCO pushes for ethical. That is like, let’s stop abiding by the minimum common denominator. Let’s try to set the bar high, because this is about the future and the present of the society. I mean, here, a steak is not like a T-shirt that I’m going to wear. It’s all my life because AI is not going to shape the way we work. It’s shaping already the way we interact, what we have access to, what we don’t have access to, how or not we are able to receive or provide services and to produce some goods, and also the relationship we have with the environment. Actually, in the recommendation of UNESCO, and again, this is going to be your night reading, perhaps, today. The recommendation is more than 100 paragraphs about it. There are two chapters of policy that we are particularly proud that countries wanted it in the recommendation. One is about ecosystem flourishing. These days, and I mean, those that program, I mean, I’ve been done in the past. Now, I don’t have the time, but let’s say, perhaps not even the capacity for the evolution that they had, but we know that by the time you learn how to code in universities, you’re told optimize the parameter, increase speed, reduce use of data, reduce points in which the algorithm gets the vote. I never remember any professor teaching me like trying to assess or assess the energy consumption. Now, this is a technology that is going to pervade each and every aspect of production and life. So by the time it goes to scale, it will matter indeed whether or not my kids, for instance, at school, are asked by the teacher, go on. whatever generative AI and create images without having any idea of the energy consumption of that just for the fun of it. And of course, we are not going to limit that, but then we have to think what is the impact of AI on the environment and the other way around, how to leverage AI for the environment. And the other chapter is something that you mentioned and that’s what you mentioned as well, Ricardo, and it’s about diversity. Diversity in terms of gender, because we know unfortunately people like me or you, we are not very common in the AI world. I should not blonde like this, it’s just because typically it’s white men, it’s a world of white men normally, unfortunately, mainly programming in English, where the presence of some languages is just marginal. So for instance, we have a program at UNESCO that is about enhancing the use and the leveraging on the Arab language. But already Arab, let’s say, is one language that was really spoken a lot. So think about the other languages that are like indigenous languages. And why we think it’s important is not only because diversity brings value, but diversity brings value in the way we actually conceive AI. That is, I don’t have the same way of going at a problem that you might have, for instance, Ash. So it means that if, by the time those of you that speak different languages know that by the time you start being fluent, you don’t translate anymore, it’s a different setting, it’s a different mindset thing. The same happens with programming in AI. So if we introduce the different programming actually it can help go in, in a way that is more enriching, accounting for more problems because then the perspectives get different. And so you check on each other and help each other. And this brings over better results. Before going to the question, let me go to the final speaker that we had. Go ahead, please.

Fadi Daou:
Thank you. Thank you very much. I think you said also that’s something that answers the question. In my perspective, I think this is a big opportunity that UNESCO put the bar higher than the compliance level, as you mentioned. If we look at our recent… the biggest damage that happened to the environment states have been done in a compliant way. And we should not repeat the same mistake. I mean, that is why we need to put the bar higher than compliance and then ethical level. However, this is our daily bread and daily work with multinational, with governments. Everybody knows how to be compliant, how to respect compliance because you have rules, you have mechanism, but almost nobody knows how to be ethical in technological production or in policymaking, because this is much more as a broader field and a very still, I would say, underdeveloped field. And this is why, as Global Ethics, we are trying to do whatever we can, in fact, to bring resources to this field. And I invite you, there is a stand in the division, you can scan the QR, you have access to our resources in this field, because we, and we have to be humble enough to say, we first don’t know how to make it ethical. We know how to make it compliant. And secondly, do we have the will? Are all the stakeholders ready to look from an ethical framework and not just a compliant framework? I’ll give you very briefly two examples. The first is the companies, the corporate sector. It’s normal that it is profit-driven, and it’s normal to bet on competition. And this comes with huge challenges when you introduce the ethical dimensions to a profit-driven sector. The second element is the political one, the police one. I am recently discussing with the United States with senior policemakers, and in China with senior policemakers. And on both sides, what they want is to protect their citizens from any harm coming from outside, from AI, external AI, but at the same time, protect their companies to be at the top level, you know, of even maybe doing harm to other citizens, but not to their own citizens. So this mentality is part of the system. So we need also, I think, to take it in a humble way to recognize the challenges that are there and then to find the solution. And I think one of the main solution is partnership. It’s like what’s happening here. I mean, have multi-stakeholder partnership that they won’t increase the profit for the corporate. They maybe won’t increase political interest on the policy-making level, but altogether they may bring more common good, I mean, for the societies and for the world. Two main, and I will finish with this, two main sectors, I mentioned them very briefly, where AI can do huge damage and increase the existing problem or can bring the solution, which is the bias. I mean, the whole bias system on gender level, cultural level, political level, whatever, AI can be a solution. But AI now is being more, I mean, based on the stuff you are trying to do, is increasing the problem itself. But it can bring the solution because it’s very simply, rather than training the machines and randomly to use the data that the data by itself is biased, to develop a way, an algorithm that can mitigate this already bias in the data, then we can mitigate and answer the problem of bias. And then the divide, the socioeconomic divide. I do believe that AI is the right opportunity for African continent, for so many other countries that can really do a leap on a socioeconomic level if we encourage AI, not just on the multinational level, but on a very small startups level where people can, we have creativity in those countries, we have innovation in those companies, we don’t have the capacity to compete with multinationals. But if multinationals can accept to mentor some startups and small initiatives, I’m sure it will bring also lots of opportunities for people all over the world.

Dr. Mariagrazia Squicciarini:
Absolutely, I mean, access is key. And this is something that I think ITU has been working for a lot of time. And actually, just for you to know, because we have here a representative of the government of Brazil. And actually one thing that we drove, let’s say UNESCO to partner with the presidency, this year presidency of the G20 of Brazil. was that they put on the table one very important thing, that is how to leverage AI to address inequalities and now to avoid that AI creates new ones. So on the table, you can see that also at the very top level of decision makers, there is this problem. Now, the topic of today is really also about information and misinformation. None of you has raised the issue that we are in a massive election year. And of course, by the time we talk about information and misinformation, that’s something that comes to mind because of course, AI can be a marvel or can work like a charm to identify fake news. But still today, and I mean, any of you that has played a bit with generative AI knows that I can put a picture of myself looking like perhaps Claudia Schiffer on a beach and nobody will be able to say whether I’m really here or I’m there. So with today, South Africa being elected their government on the 2nd of June, Mexico, for instance, and then on the 40th, if I’m not mistaken, there is India revealing the results of these six weeks, I think, in these elections. Sealed forever. Yeah. Sealed forever. Actually, there are fantastic videos about how that is deployed in a country of that size and that territory. So the question is, how do you think the relationship goes? Again, the pros and the cons. Very quickly, a couple of minutes each. I will start this time perhaps from Fadi, given that he was the last one. So the disinformation, misinformation relationship with also the democratic processes. And then we will go last for the question. Please.

Fadi Daou:
That’s a major question, of course. As you mentioned, this year is this big election year. We are organizing in July. I don’t know if you can hear me. It’s on our website, a conference with TUM in Munich on AI and human rights. And of course, we’re saying they can’t hear. Sorry, you are not hearing. Okay. So I’ll try to be a bit louder.

Dr. Mariagrazia Squicciarini:
Talk quickly, I can hear you. Yeah, because most of the people are on their cell.

Fadi Daou:
So it’s definitely a major question, disinformation. And that’s why I think the answer to this question of disinformation.

Dr. Mariagrazia Squicciarini:
Okay. a solution, we have a solution, I’m here.

Fadi Daou:
I will try also to contribute to the solution. So I think this is a major question, of course, misinformation, especially its political impact, democratic impact. Next month in Munich we have a conference on AI and human rights to tackle this question. But I think the first answer is literacy, is how to educate people on this, and that’s why next week we have here in Geneva AI and the higher education leadership summit happening here. So lots of things are happening. The question remains how to transform this policy thinking, I would say, into real policymaking and frameworks, legal frameworks, somehow to protect people from misinformation and also to leverage the capacity of people through AI literacy and digital literacy in general. We absolutely need, again, I would say, partnership, because neither governments nor civil society alone can do this. The private sector has to be part of this collaboration.

Ashutosh Chadha:
So I’ll pick up on what you talked about and the question that you pushed on misinformation, disinformation. Like before, I think there’s a technological solution and there is a policy level solution. Some of us may be aware that a couple of months ago at the Munich security conference, actually there was this coming together of organizations, and Microsoft thankfully was at the center of it, where we came up with this tech accord. And the idea of that tech accord was basically to bring together all companies to say that in this year of elections, we need to come together, bandy together, and inform people about the right ways to use AI so that it’s not used for political misadventures, if I use that statement. And I think that’s the first major step. Of course, the governments are now coming together, et cetera, but it requires a huge amount of of work on the side of both the technical companies, it requires a civil society to come together to identify how can we be proactively engaging together to prevent this technology from being used to spread misinformation and disinformation. And then there’s the technical side of it. And the technical side of it, again, spreads across multiple areas. There is the fact that you prevent this technology from being used to create such stuff. The second thing is, are we licensing this technology to organizations or players in the institution who may use it for bad purposes? The second is when the technology itself creates something, is it checking some? Is it checking whether this is wrong? And for some of us, you can actually go out and do this and try this. All of you would have some AI creator on your phone right now. You can actually, and I have tried this, I’ve actually used that and I tried to create, I gave it an option saying, can you create the picture of a woman slapping a young boy? It refused to create that. And that’s because it’s been inbuilt. Now, if by some prompt, you can actually get through it, when it creates something, it again checks it. So there are technical solutions. So I think, but the most important is to further to your point is, we need to get the governments together. We need to get the companies together. And that’s where Microsoft was at the center, as far as the tech core sector was concerned. I’ll also come to a point which you raised initially, which was very briefly, which was the ecosystem. And I think this is extremely important. It’s important that AI does not remain in the ream of the rich companies, countries or individuals. You need to create an ecosystem where it can be vibrant and which can leverage this, which is why, for example, we’ve come up with what we call our access principles. Basically. access principles allow for any company, any organization to leverage our technology, leverage our APIs and make use of our technology and sell it the way they would like to sell it or use it the way they would like to use it within the guardrails, right? So it’s access, fairness and doing it responsibly, which comes to your last point, and responsibly and ethically, obviously, which comes to your last point is that when you use this technology, you are using energy. So can you make sure that at the back end, all the data centers that you’re using or creating, are they sustainable in nature? So I think this is a journey. And like you very rightly said, we can’t say that we have all the answers. We don’t have all the answers. International organizations don’t have all the answers. Governments don’t have all the answers. We need to be humble about that. And finally, we need to be inclusive.

Dr. Mariagrazia Squicciarini:
Amanda, over to you.

Amanda Leal:
Okay. So I’m very passionate about this topic. I’ll try to be brief. And I want to focus on disinformation and democratic processes, which was your question. I think this is a huge challenge. If you think like what’s the biggest challenge, we all think of generative AI. And I want to find out, so I want to echo what Ricardo said about the fear of AI. And I want to ask you, do you feel empowered as an individual to demand information, to seek redress when you see outputs that are not ethical, that are wrong and potentially harmful? I feel that that’s why I try to focus on accountability. And I think there’s ways, if we think of an election year, I think we need to work in an iterative process. We need to deploy some institutional solutions. We need to try to work together. Because accountability is about different, how do we put this platform together for civil society as well? So we can feel empowered and then we can trust technology because there’s fear, because there’s lack of trust. And I think we won’t solve lack of trust if we first… don’t solve information asymmetry because first we don’t have access to how this this technology works and I think then that’s why I believe it’s more than access and I do recognize that are interesting interesting technical solutions coming from the industry working with academia as well but I think it’s important that we solve the information asymmetry and that’s the only way that we’re going to have inclusive participation and with an election year maybe we could deploy some platforms for accountability where we engage different stakeholders focused on this issue and finally I just want to commend UNESCO’s work with public awareness and literacy because we won’t manage to engage people if there’s no public awareness and literacy and UNESCO does like a huge work with the capacity building because we also need capacity building in the public sector the judicial sector to uphold the rule of law and youth engagement and media and information literacy so I think right now in terms of urgency we should deploy and iterate on platforms that could advance accountability but this needs to come hand-in-hand with regulation because it would unlock for example public funding with because there’s no incentives otherwise so we need public funding to advance accountability and I believe regulation is the way forward to make sure that this happens

Dr. Mariagrazia Squicciarini:
Last word to Ricardo and then what I will do is actually we’ll stand outside here we’ll take some questions if the speakers allow because I think I have the presence of our organizing in the back oh we can take five minutes extra great there you go questions thank you so much we can take a few minutes more please go

Ricardo Baptista Leite:
ahead so this extremely important topic so I healthy I we launched our community of practice in the Economist some weeks ago so if your institutions would like to join you can apply where these topics certainly will be discussed and also with the unite parliamentarians network with four billion people going to the ballot this is an issue of extreme importance for us and we’re working on it. And in healthcare, I’ll give you an example. One of the most destructive effects of disinformation was, as you all may know, the dissemination of fake news related to autism, related to the measles vaccine in the beginning of the last decade. There was a published report or article in a peer-reviewed journal of public health in 2018 that looked at those social media posts that were disseminating that information between 2014 and 2017. And over 90% of the original posts were from farm bots, bot farms in the Russian Federation. And so I think we need to acknowledge that there are bad actors out there. And that if we do not address the bad actors that are disseminating purposefully this kind of misinformation, that this will not be solved. And why do bad actors decide to hit something like vaccination? It is because vaccination is one of the areas that focuses between the trust and depends on the trust between the state and the citizen. If you’re able to disrupt that link of trust, you are able to have ripple effects that we have seen throughout the years now when it comes to the vaccination issue. And so this is undermining a series of efforts we are making, particularly when it comes to the power of multilateralism. And I would say that in politics, we are also facing another challenge that is related to misinformation, which is we’ve lost the sense of shared reality. Before one could say that the saying the truth was not necessarily the mainstream way of doing politics. And since Richard Nixon, we can remember that. But now we have a problem of understanding the basic concepts of the reality we live in. So we have a table in front of us, but someone can come into the room. the room saying, this is not a table, this is a horse. And the conversation becomes absolutely impossible if you do not share the reality. And that’s the way we are seeing this kind of disinformation spreading because that’s where artificial intelligence, and Fadi was mentioning, it can be the greatest equalizer, but it also has a risk of becoming the greatest divider if we do not get the ecosystem right from the start. And the power that AI has, the destructive power that it has needs to be harnessed and controlled from the start also so that we can make sure that the AI is used for good. And so I just like to end by saying on this topic, it’s important for companies to have this conscience, which I think is growing in terms of the ethical use of AI, but also in the educational sector, we’re still teaching students as if we were in the 19th century. We don’t need them to memorize the way we did in the past. What we need is critical thinking. We need people not to take whatever they read as reality, but actually to question everything in society and everything they are faced with, especially with the rise of AI, and to be able to be a counter to attempts to disseminate disintegration of what has been the strong points of connecting us as society and as democratic systems. And if we are unable to do that, I think democracy can be at risk. And so I think it’s one of the greatest challenges of our time.

Dr. Mariagrazia Squicciarini:
Excellent. Actually, Ananda and Ricardo touched upon an issue that is actually related. And it’s a lot of information asymmetry that perhaps to some doesn’t say anything that means I have information that you don’t have and I can leverage it for the purpose. And you’re talking about the next level of the game that AI has to it, that the creation of information, additional perhaps of having on any information that others don’t have, that I can already edge upon. Then I can create my own, at times, fake information and spread it, the horse example, so that I have the bargaining position. And also with you, the other, I have a completely different position. because in the inability, and this is what you said, of discriminating between what is right and what is wrong and what is real and what it is not, of course, people take things at face value. So hence, going back also to Fadi, the issue of raising awareness, and this is something also that Amanda and Ashu mentioned. Now, I have three questions. I have seen three hands raised. So please, say briefly your name, very quickly the question, and so we go to the answers. Why more quickly than others?

Audience:
My name is Boris Engelson. I am a jury freelance and accidentally president of the Commission of Ethics of my organization. By the way, I am the only jury attending from the first session of WSIS 20 years back. This is already telling. Why don’t others come? Because precisely among the thousands, the millions of juries on earth, we could never come agree on a single item relating to ethics or information or whatever. I’m very impressed to find at each session of WSIS, hundreds of professionals who are so clear about what is good and bad, what is right and wrong. We journalists cannot to the point that we just don’t attend. Now, I have been very much interested at your opposing, or let’s say, making the difference between ethics and responsibility. I have seen lately a film making the difference between justice and fairness. That would be a lengthy discussion, but ethics by essence is a solitary minority position. When it becomes a norm, it’s just professional, careerism, and let’s say law. Ethics, we better don’t talk about that. You refer to UNESCO, which is not UNESCO. UNESCO can possibly tell what is true and intelligent in science, surely not. education, surely not in culture, and surely not in ethics. So, I have seen also from the start that here you write information society with high case. I do not trust your answers, of course. It is just to tell you that information society with high case, I discuss here, is just a tiny part of information society in the general sense. And if I have to go and find for natural or artificial intelligence, I will go outside this building, surely not here.

Dr. Mariagrazia Squicciarini:
Okay, so I’m very glad to hear, knowing that in 45 minutes the ability of dealing fully with a problem, we will be getting all the Nobel Prize. And I mean, everybody, this is a free country and a free place to express, is free to decide what they like or what they don’t like. Now, ethics, as I said, is human rights, human dignity is a fundamental freedom, so it’s not a philosophical concept. But insofar, it is very close.

Audience:
This is a question. Great. Okay, the solution was, we’ve got to talk to companies like Microsoft. Microsoft, in case you don’t know it, is preparing to release VASA 1. VASA 1 is a facility that anybody can buy, that with a single still picture and 30 seconds of speech, you can make it say anything. And they are proud that facial things and lip coordination works. And they assert, as one of its values, that it contributes that the ability for changing facial recognition and direction contributes to the perception of authenticity. That is, it facilitates lying. And they’re going to send it out. And you say we should be dealing with Microsoft. How do we handle that kind of nonsense?

Dr. Mariagrazia Squicciarini:
Well, I’ll go to Microsoft very quickly because I have the lady that wants to, actually, you know, please ask the question so that then we have the answer right in a row, because we don’t have time. We have to be out in two minutes.

Audience:
I just wish to point out, everybody referring to AI. technology, in the singular, is plurals. It’s facial recognitions. Actually, the worst is not having your name not recognised, but the fact you know all your DNA and exactly what to test you for and what to find you for. So it’s just you have to be talking about AI technologies in the plural, because otherwise it’s just not.

Amanda Leal:
Well, we can talk interchangeably, liberty or liberties. First, the duty of the answer. Just for clarity about AI technologies, what I identified in our conversation here is that what Ricardo brought, for example, is related to recommender systems, how this information is propagated and amplified. And Ashu talked about generative AI tools, and I talked about the impact of generative AI in the information ecosystem. So that’s a good observation. But for clarity, I think we’re more in the realm of looking at these two aspects.

Dr. Mariagrazia Squicciarini:
Absolutely. We are for plural inclusivity. The last word to Ashu.

Ashutosh Chadha:
Yeah, and I’m very glad that you’ve asked that question, because these are exactly the questions that need to be asked when we create technology, which has both a potential to do good and a potential to do harm.

Audience:
There’s profit in it. That’s why they’re doing it.

Ashutosh Chadha:
While there’s going to be profit, there are also people who are like bad actors who will anyway use this technology to create bad stuff. What is incumbent upon all of us is that we create those connects as well as those gates within the system so that it can’t be used for the bad harm that is. The example I gave you, that if you go into, for example, a creator thing right now and say, create the picture of a woman slapping a boy, it will not create. Create the picture of possibly a politician saying something wrong, it will possibly stop. The same rules need to be applied to any technology that happens. Will we control it at the first instant? Possibly not, but it’s a learning experience and everybody is working towards that. You can’t throw the baby right now out of the bathwater, you have to make sure that you build the guardrails.

Dr. Mariagrazia Squicciarini:
Thanks a lot for everybody.

Audience:
You will get the understanding you’ve been talking about on the internet, but I can also give it to you by email and I will send it to you.