Kseniia Fontaine, Neil Sahota, and Frederic Werner on AI for Good

Kseniia Fontaine is Senior Business Development and Operations Officer in ITU’s Strategic Engagement Department, supporting AI for Good and emerging-technology initiatives. She advances partnerships, resource mobilization, standards engagement, skills development, and inclusive capacity-building, helping governments, industry, academia, and civil society expand equitable access to responsible artificial intelligence globally.

Frederic Werner is Chief of Strategic Engagement at ITU and Chief of Strategy and Operations for AI for Good. Based in Geneva, he helped create the AI for Good Global Summit and leads initiatives connecting governments, researchers, industry, and civil society to practical AI applications serving sustainable development.

Neil Sahota is loanDNA’s Chief AI Officer. Also, he is now the author of second book, AI Activation Code: https://aiactivationcode.com/. He is a United Nations AI Advisor, IBM Master Inventor, UC Irvine professor, and author of Own the A.I. Revolution. With more than two decades of business experience, he helps organizations develop AI strategy, enter new markets, and build emerging-technology solutions.

Scott Douglas Jacobsen, Frederic Werner, Neil Sahota, and Kseniia Fontaine discuss AI for Good’s origins, governance, standards, human rights, skills, mental health, misinformation, space, healthcare, Africa, and inclusion, stressing anticipatory ethics, practical collaboration, and public engagement to ensure AI serves humanity rather than deepening global digital exclusion worldwide.

Scott Douglas Jacobsen: How do you define AI for Good? Who would prefer to start?

Frederic Werner: Yes, I can answer how we define AI for Good. I will start with a short story. There is an AI safety and security expert, Roman Yampolskiy, who has been on several podcasts recently, including the Lex Fridman Podcast. I met him in New York, and he asked me an interesting question. He said, “Fred, is AI for good, or for good?” I asked what he meant. He said, “Is it for good as in beneficial, or for good as in forever?” I thought it was a clever play on words, but then he clarified: what if AI is the last thing humans ever invent?

That reframing stayed with me. You may agree or disagree, but it is not difficult to imagine a future where AI is responsible for many new inventions. That is already beginning to happen, or at least human creations are increasingly co-created with AI. If that is the direction we are heading, then we need to ensure that AI is for good in the sense of being beneficial.

One of our key phrases for the summit is: what is AI for Good? Simply put, it is about unlocking AI’s potential to serve humanity.

Rewinding a bit, Neil was present at its inception in 2017. That is when AI for Good was established by the International Telecommunication Union, the United Nations agency for digital technologies. At the time, the conversation around AI was dominated by hype, fear, and uncertainty. We had to persuade people to come into the room. It was not for lack of strong speakers, we had leading figures such as Yoshua Bengio, Fei-Fei Li, Stuart Russell, and Gary Marcus, but at that time they were not widely known outside expert circles.

We were trying to show people that, while there was hype and fear, there was also real promise. If there was even a chance that AI could develop in this way, then we needed to ensure it was used to address global challenges, affordable healthcare, clean energy, food security, disaster response, and beyond.

Misinformation was a concern, but the potential was there. At the time, the mission was simple: identify practical applications of AI that could help advance solutions, and then build collaboration around them within the UN context. When I say collaboration, I do not mean only big tech, major governments, or a select few with funding and resources. The goal was to bring the Global South to the table, along with students, young people, artists, athletes, leaders, NGOs, civil society, politicians, international media like yourselves, and everyone in between. I will pause there, that is the genesis of AI for Good. But if AI is going to be “for good” in the sense of lasting, then we must ensure it is genuinely beneficial.

Neil Sahota: To add to that, one reason I championed this initiative was the amount of fear and fear-mongering at the time. Bad actors will always do harmful things, but good actors often need a nudge. That is what we wanted to create. People tend to focus on profit or entertainment, not that those are inherently bad, but we do not always think about how to use technology for broader benefit, such as advancing development goals. We tend to separate for-profit and nonprofit efforts, even though they can coexist and even complement each other.

The second point is that I do not think AI will be the last thing humans invent. AI can only do what we train it to do, and we have not yet figured out how to teach genuine creative thinking. Perhaps that will change. For now, I strongly support the idea of hybrid intelligence, humans working with machines. That is where the real benefit lies. I do believe AI is making us more human.

Kseniia Fontaine: Building on that, when we brought AI experts to the table, many of them emphasized that AI is too important to be left only to technical specialists. For the International Telecommunication Union, as a UN agency, we approach this practically by grounding AI for Good in established UN principles. When we develop projects and build the community, we follow human rights frameworks that are not defined by one or two countries but represent a global consensus on the challenges the world is facing. These frameworks serve as our benchmark. We are not inventing new ethical foundations; we are building on internationally agreed principles that have developed over many decades.

Jacobsen: There is a well-known quote, if I recall correctly, from Alan Turing, which can be interpreted as either realistic or pessimistic. It states that it is customary to claim that some uniquely human characteristic can never be imitated by a machine, but that no such limits can confidently be set. If that is the case, then in thinking about AI for Good, are we essentially asking the same question we ask about human beings? That is, how do we orient both people and machines toward what we define as the good, especially if no clear boundaries exist?

Sahota: That is an interesting point. Ultimately, it comes down to people. We are getting philosophical, but “the good” is a subjective concept. We have already seen troubling uses of this technology. For example, there was a case in Canada where deepfake technology was used to create child exploitation material. That is precisely why AI for Good is necessary, not just involving AI experts, but bringing everyone to the table. We need to become better at anticipating misuse. There is an entire initiative within AI for Good focused on governance for this reason. Technology is engineered toward intended outcomes. We build what we are asked to build, but we do not always anticipate how it might be misused.

I think it is a fair assumption that other people will figure out harmful uses, and that is one of the major challenges. Historically, regulation has reacted after something goes wrong. With AI, things are moving so quickly that we cannot simply react anymore; we have to anticipate. Humans are not very good at that. Ironically, AI is quite good at scenario planning. Still, putting yourself in the mindset of a criminal, a bad actor, or someone seeking power or money is difficult for most people.

What we are trying to figure out is that, ultimately, there will need to be some ethical baselines the world can agree on to define what “good” really means. That is the difficult challenge. What we are discovering is that the hardest part is not making the technology work, or even developing applications. It is the people problem. That is what we are trying to address through this community. If we bring good people together, and I believe most people are good, we can accomplish a great deal. But we need an ecosystem, a hub, a forum, and a nudge to create the good.

Werner: Responding to that quote as well, one interesting thing about AI is that it acts almost like a mirror. It pushes us into a corner and forces us to think about things we have not naturally thought about for a long time. I would even argue there has probably not been a time in history, apart from perhaps the Greek or Roman philosophers, when we thought so much about what it means to be human.

As these systems evolve and become more capable, every new capability forces us to ask: What would I do? What is right? What is wrong? What is ethical? Of course, there is the classic trolley problem in autonomous driving, but that is only one example among many. As these capabilities evolve, humans need to think about rules. Is it safe? Is it ethical? Is it inclusive? Will it serve everyone or only a few? Will it respect privacy? Will it cause harm?

It almost forces us into a corner where we have to address these questions. What is interesting is that when we see a piece of software or an autonomous vehicle, for example, I was recently in the United States and visited Waymo. I was impressed, frankly, driving around Austin. I do not have the statistics in front of me, but from what I understand, the data on hours driven versus accidents, especially involving pedestrians and cyclists, is quite promising, though it still needs to scale. It would not work the same way in New Delhi, where I had just been.

Still, people often seem more concerned about the idea of a machine making a mistake than about humans making mistakes. I was half-joking, but only half-joking, when I said maybe we need a “Humans for Good” conference. Sometimes it feels as though the focus becomes, “Let us talk about all the machines going wrong,” when human behaviour remains the deeper issue.

Fontaine: Yesterday, we had a very interesting discussion with the Holy See Mission in Geneva. Last year, we received a message from Pope Francis, and this year it looks as though the Holy See may return to Geneva, because this topic has become very important. We have found that the Vatican has been deeply involved in philosophical discussions about technology.

One point they raised was that, because of these questions around the development of AI, they see more people turning back to religion, or at least turning again toward questions about what it means to be human, what values matter, and how technology should be used.

There are also new uses of AI, such as AI for grief. There are now applications that allow people to recreate deceased loved ones, using their image or voice as part of the grieving process. Is that good or bad? We do not know. Similar questions arise around longevity and life after death. There is also Kazuo Ishiguro’s Klara and the Sun, which presents a dystopian reflection on recreating or replacing children in the context of illness and loss. These questions are returning with urgency.

So, you cannot measure good or bad with one code or one rule. But as Frederic said, AI makes us question ourselves much more deeply as human beings. It is not only about whether the technology itself is harmful; it is about how we use it. As Neil said, it can be used for the wrong purposes. That is why it is important to have a community driving AI for Good, not only from our side, but through a broader community that questions these applications in real time as they develop. That is very significant. So maybe we really do need “AI for Human.”

Jacobsen: The thrust of the Turing quote was the convergence of machine capabilities, not simply intelligence, with human capabilities, which change much more slowly, over something like a 100,000-year timeline of evolution and speciation. So, an AI for Good summit, or part of it, is in some sense also an “Humans for Good” conference.

Because we are talking about convergence, which we are seeing more and more now, we are also talking about redefining the good. We are discovering more about human nature through machine intelligence, while also, to Neil’s earlier point, redefining human nature. There is a circularity here: human nature is natural, but it is also adaptive by nature.

Maybe that helps explain why people reach for new answers in older traditions, or reinterpretations of older traditions, to define the good. What do you think are the most immediate concerns for communities trying to develop anticipatory legislation or moral frameworks for this rapidly advancing curve of artificial intelligence?

Sahota: I will jump in. One of my biggest concerns is that, because we are teaching AI, people are relying on technologists to build things they do not fully understand. We cannot do that successfully without having the right people at the table.

I will use Meta as an example. They got into trouble over redlining, which in the United States refers to practices that can exclude certain groups from housing-related opportunities or advertising. They had built AI tools to facilitate advertising, but when people asked their engineers how they accounted for redlining, the engineers apparently did not know what redlining was. That is the issue.

A lot of people still treat AI as an IT project: gather the requirements, build the system, deploy it. But that is not really how AI works. I often say you are not buying software; you are hiring a high-energy intern. It is ready to do work, but you have to teach it how to do that work, your processes, your data, your context.

Take legal tech. If I remember the numbers correctly, venture capital and private equity firms invested heavily in AI and legal startups over several years, and many of them failed. But when you look at who survived, the successful companies were often started by lawyers or paralegals. They understood the domain. They were not just building software; they were effectively building an AI paralegal or an AI associate lawyer.

That is where we struggle. We have siloed AI as technology, so people assume it belongs to technologists. But that is not true. Would you entrust the hiring of a financial analyst to your data scientist? Probably not. Not everyone needs to know how to make the hammer, but people do need to learn how to use it. That is the challenge. We have this new tool with unfamiliar capabilities, and too often we default to the help-desk person to show us how to use it, instead of learning how to manage it ourselves.

Werner: That is a good example, Neil. Another example applies more to the developing world. We held AI for Good Impact Africa at the end of last year. The idea was that not everyone can come to Geneva to discuss these questions, so we also need to bring these conversations to different parts of the world.

There was a major discussion around AI diffusion, sovereign AI, and sovereign data. Six or seven years ago, before the pandemic, some large technology companies had a mentality that was almost: “Here is a technology handout. It works out of the box. Use it, trust us, and everything will be fine.” That may have seemed acceptable before the pandemic, but now there has been a shift.

I would compare it to what happened in East Africa with mobile payments. There was a mobile-payment revolution that effectively leapfrogged decades of legacy infrastructure. With systems such as M-Pesa, countries were able to reduce dependence on intermediaries, improve financial inclusion, and generate wider beneficial effects. There is hope that AI could do something similar on the African continent by unlocking new opportunities.

I was speaking with a gentleman in Johannesburg who works in the townships and was writing a book on AI in the townships. I asked him how many people in the townships have a mobile phone with 5G and access to ChatGPT or something similar. He said probably about 80 percent. That is a significant shift compared to a couple of decades ago.

Then we started discussing whether people are using these tools to start businesses, create value, collaborate, educate themselves, or ensure that people are not left behind. That brought us to AI skills and education. Putting a tool in someone’s hand does not mean it will automatically be used in the right way, or that it will create value and new businesses responsibly.

So, the AI-skilling and education piece, from grade school to graduate school, from children to diplomats to entrepreneurs and everyone in between, is probably the top priority. If you speak with our member states who come to AI for Good each year, they discuss many major issues and do not agree on everything. But one thing they agree on across the board is the need to address the AI skills gap and democratize access to AI skills. That is a shared challenge whether you are in the United States, South Africa, China, Bangladesh, or elsewhere. As Neil said, just because you have the hammer does not mean you know how to use it properly.

Fontaine: We have already been trying to address this issue. Last year, we launched the AI Skills Coalition, a global movement led by ITU with more than 70 partners worldwide, including UN entities, private-sector organizations, and NGOs. The idea is very basic: democratize access to AI skills, covering everything from K–12 curricula in different parts of the world to diplomats and policymakers.

Even when we work with highly professional people, we sometimes see that their understanding of AI is still very basic, even though they are making decisions about it. The same applies to schools. In the Africa region, for example, we have been running a project focused on AI and robotics in schools, working to localize curricula through governments.

There is a great deal to do and a great deal of demand. The question now is pace: how quickly can we move? There is also the challenge that the AI skills needed in Europe may not be the same as those needed in Africa or Asia. So we need partners who can localize knowledge. Global frameworks are useful, but we must ensure they are applicable in specific regional contexts.

Jacobsen: One aspect of this is the democratization of the skill set. But use will be local, specific, and sometimes parochial. Another aspect is human rights. The aspirational and foundational principle behind human rights is universalism, and that part is not necessarily solved. That seems to be one of the questions AI in Africa is posing.

There has always been a challenge in applying universal human rights as an ethical framework. How do you see similar problems potentially arising when applying a reasonable universal definition of “good,” within an AI framework, to different regions of the world with different needs?

Werner: One very important development this year is that, for the first time, the United Nations AI Global Governance Dialogue will take place in Geneva at the same time as AI for Good.

That is significant because AI for Good has existed since 2017 and has grown organically as a movement and community. During that same period, ITU has also been leading and fostering AI governance coordination within the UN system. We co-lead the interagency working group on AI with UNESCO.

Out of this momentum, there have been several landmark resolutions passed in New York. One of them established the first Global Governance Dialogue in Geneva as part of AI for Good. This will be the first time member states sit around the table in a mandated way to discuss AI governance and how to move it forward.

It will work in several ways. A scientific panel on AI was nominated and announced a few months ago. ITU was partly responsible for selecting that panel, together with UNESCO and the Office of the Secretary-General’s Envoy on Technology in New York. What is interesting is that probably about a quarter of the nominees have already been involved in AI for Good from the beginning, including Yoshua Bengio, for example.

There are consultations happening now in New York, Geneva, virtually, and around the world. Member states will come to the table for consultations and dialogue over two days. The scientific panel on AI will deliver a report to the dialogue. The output, without sounding too bureaucratic, will be a report from the co-chairs of the dialogue, the ambassadors of El Salvador and Estonia.

What is significant is that the dialogue has seven themes, and one of them is human rights. Human rights will therefore be central to member-state discussions in an organized way. That will feed into the global summit and into different mechanisms across the UN system.

Our goal is to turn principles into action. There are many high-level principles emerging from resolutions, declarations, reports, and regulatory processes around the world. One of the most effective ways to translate those principles into action is through standards. At its core, the International Telecommunication Union is a standards-setting body. It is the only UN agency with a mixed membership model, 194 member states, along with more than a thousand private companies and academic institutions, all contributing to a consensus-driven standards process.

I do not think we will see a centralized global AI regulation. AI is a moving target, and different regions and nations have their own priorities. But standards allow us to take high-level principles and operationalize them. The ITU standards-making process is inclusive by design. There are significant efforts to bring developing countries to the table through training and funding programs so they can participate. That way, standards are not just designed to work in places like Los Angeles or Shenzhen, but are fit for purpose across regions such as Africa.

Human rights and fundamental rights are built into this process. As standards are co-designed, by member states, companies, experts, and academics from around the world, there is an opportunity to embed considerations like security, inclusivity, sustainability, energy use, and privacy. Standards may not be the most glamorous tool, but while we wait for a comprehensive regulatory solution, which may never fully emerge, they are the most practical way to embed “AI for Good” principles as systems scale.

Sahota: I will add that what Fred is saying is incredibly important. I have often been asked, “Why the ITU? Why the United Nations?” The honest answer is that they are the only globally credible organizations positioned to do this.

Before AI governance gained traction, there were initiatives like the Partnership on AI, but many were launched by private companies, which led people to question their motives. Others were driven by individual states, raising concerns about political agendas. That lack of trust was a major barrier. This brings us back to the point that people are often the biggest challenge.

The ITU and the UN stepped in to provide a neutral platform and have done a strong job driving this forward. When I look back over the last decade, the progress is significant. The conversation around AI has changed dramatically. Ten years ago, many discussions focused on fears that AI would turn against humanity. We do not hear those narratives as much anymore. Instead, people recognize AI as a powerful tool.

Even during the pandemic, when relations between the United States and China were strained, AI governance remained an area of continued dialogue. That shows how important this issue is globally. Work around standards and governance is critical because it surfaces second-order effects we do not always consider, for example, how AI infrastructure drives demand for rare earth materials, potentially creating new geopolitical and economic dynamics. For developing countries, that raises serious questions about participation and equity. Some African nations are already collaborating to address these challenges collectively.

That is why we need this community, to anticipate consequences, understand potential uses and misuses, and guide development toward positive outcomes.

Jacobsen: Thank you very much for the opportunity and your time, everyone.

Scott Douglas Jacobsen is a blogger on Vocal with more than 200 publications on the platform. He is the Founder and Publisher of In-Sight Publishing (ISBN: 978–1–0692343; 978–1–0673505) and Editor-in-Chief of In-Sight: Interviews (ISSN: 2369–6885). He writes for International Policy Digest (ISSN: 2332–9416), The Humanist (Print: ISSN, 0018–7399; Online: ISSN, 2163–3576), Basic Income Earth Network (UK Registered Charity 1177066), Humanist Perspectives (ISSN: 1719–6337), A Further Inquiry (SubStack), Vocal, Medium, The Good Men Project, The New Enlightenment Project, The Washington Outsider, rabble.ca, and other media. His bibliography index can be found via the Jacobsen Bank at In-Sight Publishing, comprising more than 10,000 articles, interviews, and republications across more than 200 outlets. He has served in national and international leadership roles within humanist and media organizations, held several academic fellowships, and currently serves on several boards. He is a member in good standing in numerous media organizations, including the Canadian Association of Journalists, PEN Canada (CRA: 88916 2541 RR0001), Reporters Without Borders (SIREN: 343 684 221/SIRET: 343 684 221 00041/EIN: 20–0708028), and others.