Uganda bureau of Statistics

For me, a successful first Global Dialogue on AI Governance would be one where we leave with more than just a communiqué or a list of good intentions. We need real commitments that people working with data on the ground can actually see and feel. At Uganda Bureau of Statistics, my work involves making sure data reaches the people who need it from policymakers to ordinary citizens. AI is already changing how we collect, process and share statistics. We are rolling out digital data collection through mobile surveys and CAPI systems, and AI governance directly affects how that data is handled, who controls it, and who benefits from it. A lot of Uganda's data ends up processed on foreign platforms that question of sovereignty cannot be an afterthought in governance conversations. Uganda already has a National Data Policy and a National AI Policy adopted in 2024. But national frameworks alone are not enough without global alignment. NSOs like UBOS feed data into SDG monitoring and national planning if AI tools distort or misrepresent those statistics, the consequences are real. So governance must answer a basic question: who validates AI-generated statistics? There is also a trust issue. Citizens already question some government data. If AI is making decisions invisibly behind our numbers, that distrust will deepen. A successful Dialogue must push for explain ability standards so that institutions like ours can communicate AI-assisted findings openly and honestly. Finally, Uganda does not yet have enough AI auditors, ethicists or specialists. Success means walking away with a concrete commitment to capacity building through South-South and North-South technical assistance so that UBOS and institutions like it can govern AI confidently, not just consume it.

• AI capacity-building
• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Transparency, accountability, and human oversight

As a data disseminator at the Uganda Bureau of Statistics, my priority selections are grounded in the day-to-day realities of working with national data in a developing country context. AI capacity-building is my foremost priority because Uganda, like many African nations, lacks sufficient AI auditors, ethicists and technical specialists. Without deliberate investment in human capacity, countries like ours will remain consumers of AI tools designed elsewhere, with little ability to shape, question or govern them independently. Safe, secure and trustworthy AI follows directly from my work at UBOS. We manage sensitive national datasets census records, household surveys, and indicators that feed directly into Uganda's development planning and SDG monitoring. Any AI system touching this data must meet clear safety and security standards. The stakes are too high for ambiguity. The social, economic, ethical, cultural, linguistic and technical implications of AI matter deeply because the communities we serve are not abstract. Uganda has over 56 langugages, largely rural, and home to populations that are often invisible in global AI datasets. Governance that ignores these realities will produce tools that either exclude or misrepresent the very people they claim to serve. Finally, transparency, accountability and human oversight are non-negotiable for an institution like UBOS whose credibility rests on public trust. data users already question some government data. If AI is generating or influencing official statistics without clear explanation or human accountability, that trust erodes further. Governance must ensure that humans not algorithms alone remain answerable for the numbers that shape national policy.

Yes, there are some important issues that I feel are not fully captured by the listed themes, and I want to raise them from where I sit - working with national data in Uganda. The first is about who really owns our data. At UBOS, we collect data about Ugandan people and communities. But a lot of that data ends up being processed or stored on foreign platforms and servers. We do not always have full control over what happens to it after that. This question of data sovereignty cuts across many of the listed themes but is not directly addressed by any one of them. African data should serve African development first. The second issue is about protecting the credibility of official statistics. NSOs like UBOS are trusted to produce the numbers that government, donors and citizens rely on. But today, AI can generate synthetic data and automated estimates that look very official. If people cannot tell the difference between AI generated figures and figures that UBOS has actually verified and validated, that is a serious problem. We need governance that protects the authority of official statistics. The third issue is about people who are hardest to count. In Uganda, many people are in the informal economy small traders, subsistence farmers, people in rural and hard-to-reach areas. Most AI systems are built around populations that are formal, connected and documented. This means AI could actually widen the data gaps we are already struggling to close, making already invisible communities even more invisible. i feel that these 3 issues do not fit neatly into one theme. But they are very real where I come from, and a Dialogue that does not make space for them will miss a big part of the picture

Here's the simplified, very human version: honestly speaking what we are seeing on the ground in Uganda. On capacity building we simply do not have enough people who understand AI deeply enough to question it or govern it. When donors or government programmes bring in AI tools, we use them. But most of us do not fully understand how they work or what can go wrong. Uganda passed a National AI Policy in 2024 which is a good start, but a policy alone does not build skills. We need real, practical support to train people inside institutions like UBOS. On safe and trustworthy AI: most AI tools we encounter were built somewhere else, using data from somewhere else. They do not always understand Uganda's languages, our geography or how our communities are structured. When these tools inform planning decisions, wrong outputs can cause real harm and nobody notices quickly enough. We are digitising our data collection at UBOS right now, so this is exactly the right time to get the governance right. On social and cultural implications : AI is already spreading information and misinformation in Uganda, including in local languages like Luganda. Most content moderation tools cannot even read these languages, let alone flag harmful content. This is a gap that affects ordinary people every day. On transparency and accountability, many Ugandans already do not fully trust government statistics. If AI starts quietly influencing official numbers without any clear explanation, that trust will get even worse. But if we get this right if we can clearly explain how data is produced and verified so this is actually a chance to rebuild public confidence in institutions like ours at UBOS

Uganda cannot govern AI alone. The technology is moving too fast. Most decisions about how AI is built are made far away from us. That is why this Dialogue matters so much to countries like ours. The first thing the Dialogue can do is make sure every country has a real seat at the table. Not just the rich and powerful ones. Uganda has a National AI Policy but we had very little say in the global frameworks that affect us most. That has to change. Cooperation must start with genuine inclusion. The second thing is connecting agreements to real action. It is not enough to sign a document with good principles. For cooperation to mean something in Uganda, it must come with practical support — training, partnerships and tools that institutions like UBOS can actually use. Agreements that stay on paper do not help us. The third thing is protecting smaller countries like our Uganda when we deal with big technology companies and foreign platforms. Right now the relationship is unequal. Our data flows out. The benefits flow elsewhere. Common international rules can give countries like Uganda more leverage and more protection. The fourth thing is creating space to learn from each other. Uganda does not need to figure everything out alone. If another African country in East Africa , or Africa has found a good way to govern AI in their national statistics system, we should be able to access that experience easily. South to South learning is often more relevant to us than advice coming from very different contexts. I am coming to this Dialogue representing uganda / UBOS (i pray i could get full sponsorship) am passionate about governance because am doing a PhD in data governance and representing a country that is willing to engage seriously. But we need the Dialogue to be equally serious about us not just as an audience, but as partners who have something real to contribute.

Strategic Foundations The Dialogue should build upon the National AI Policy (2024) and the National Data Policy. These, alongside the UBOS Statistical Act, provide the legal guardrails for data handling. Regionally, it must align with the African Union's AI Continental Strategy and East African Community digital governance initiatives. This ensures Africa is a lead protagonist in the conversation, not just a subject of it. Existing Partnerships Uganda already collaborates with the UN Statistics Division and the World Bank on data modernization. Key local initiatives include: Makerere's AI Research Cloud: Providing high-speed GPU "brain power" for complex testing. Sunbird AI: Developing practical, open-source tools for local language translation and public transport. AI Health Lab: Using AI for faster diagnoses and maternal health tracking. Otic Foundation: Training citizens in cities like Kabale through the "AI in Every City" program. The Added Value The primary value the AI Dialogue brings is cohesion. Currently, partnerships are scattered. Uganda's policies do not always align with global standards, creating regulatory gaps that platforms exploit. The Dialogue can: Bridge the Gaps: Harmonize domestic rules with international frameworks so they "talk" to each other. Enforce Accountability: Use the diplomatic weight of a UN-led process to move beyond "feel-good" documents toward binding commitments from tech giants. Localize Innovation: Ensure homegrown tools, like the Ministry of ICT's "Sunflower" AI model, are supported by global infrastructure. We have enough talk; Uganda needs a Dialogue that produces functional, ready-to-use tools for social good.

Everyone has something to bring to this conversation. But only if the conversation is set up in a way that lets them speak. Governments like Uganda's can share what is actually happening on the ground. What policies we have. What is working. What is not. We should not just be invited to listen to others — we should be asked to present our own experiences. Institutions like UBOS can bring real data and real stories. We see how AI affects statistics, planning and public trust every day. That kind of practical knowledge is very valuable in a room full of policy language. Technology companies should be at the table too. But they should be asked hard questions, not just given space to present their products. They need to explain how their tools work in low-resource settings and what happens when things go wrong. Civil society and community groups should be there to represent ordinary people. Especially people who are not online and who have never heard the words artificial intelligence but whose lives are already being affected by it. Young people should have a dedicated space. They are growing up with this technology. Their voice matters. For the format, I would recommend small working groups where people can speak freely, not just big plenary sessions where only confident speakers get heard. There should be translation into local languages. And there should be a clear way for what is said in the room to actually influence the final outcomes. Too many dialogues collect our views and then ignore them

People from rural communities are missing. In Uganda, most people do not live in Kampala. They live in villages. They farm. They trade in local markets. AI is already affecting their access to loans, health information and government services. But nobody is asking them what they think about it. Women and girls are underrepresented. Especially women who are not highly educated or not working in tech. AI systems are often built without thinking about how they affect women differently. People who speak local languages are missing. Uganda has over 40 languages. Most AI tools only work well in English. The people most affected by that gap are not in these governance conversations. Small institutions like district local governments, community health workers and village level data collectors are also absent. They work with data every day but they are never consulted. To include these voices, the Dialogue needs to do more than send an online form. It needs community consultation meetings in local languages. It needs organisations like UBOS to go out and collect views from the field before arriving in New York or Geneva. It also needs to stop assuming that participation means flying someone to a conference. Virtual participation, translated materials and local dialogue forums can bring in voices that would otherwise never be heard.

I thing the dialogue can try out different engagement formats like - Use of Live case studies. Instead of presentations, bring real examples from different countries and ask the room to work through them together. What went wrong. What could be done differently. Learning from real situations is much more useful than listening to theory. Open microphone sessions for smaller voices. Set aside time specifically for people from developing countries, civil society and young people to speak without having to compete with powerful delegations for the floor. Parallel community dialogues. While the main Dialogue happens at the global level, partner organisations in different countries should be running local conversations at the same time. The findings from those local conversations should feed directly into the global one in real time. Digital participation walls. Use simple technology to let people who cannot travel share their views during the Dialogue itself. Not just before or after or during. So their voices are heard while decisions are still being shaped. Follow up sessions. The worst thing about most global dialogues is that they end and nothing happens. Build in follow up meetings every six months where progress is reported back honestly. For me personally, the most important thing is that the format respects people's time and produces something real. People in Uganda are busy. If we take time to engage, we want to know it made a difference. A Dialogue that listens but does not act is worse than no Dialogue at all.

First, Uganda's own National AI Policy of 2024 is something I am proud of. It is not perfect but it shows that we can sit down as a country and decide what we want from this technology. It talks about ethics, about data protection and about making sure AI serves Ugandan development goals. More countries need to do this. And those of us who have done it need support to implement it properly. Second, the work that UBOS is doing to digitise data collection and dissemination is a practical example of governing AI from the inside. We have worked hard to make sure our statistics reach the right people the policymakers, researchers, journalists and ordinary citizens through our website, reports and data portals. But as AI starts to shape how data is packaged, summarised and shared, we have to ask harder questions. Who is deciding what gets highlighted? What gets left out? AI in data dissemination can simplify access but it can also quietly distort what people see and what decisions they make. Governance must cover this. Third, the African Union's data policy framework talks about data sovereignty and making sure Africa benefits from its own data. That is the right conversation. Fourth, Rwanda has done interesting work on AI regulation and digital governance. As an East African neighbour their experience is more relevant to us than frameworks from Europe. We need more of that regional learning. What I would tell this Dialogue is simple. The best governance is not the most complicated. It is the most honest. Tell people what AI is doing with their data. Make sure official statistics cannot be altered or misrepresented by automated tools without human verification. And when something goes wrong, make sure someone is actually responsible. That is what good governance looks like from where I sit