AWE

For me, success would come down to three things, and I say this as someone who works in data analytics, portfolio governance and is currently studying AI and Data Analytics on a post graduate level. I am someone who uses these tools at work and is actively trying to understand the responsibilities that come with that. First, agreement on a shared baseline that cuts through the current fragmentation. Right now, organisations and countries are approaching AI governance very differently, which creates real challenges for anyone trying to do the right thing, because the standards keep shifting and there is no common floor. A successful first Dialogue does not need a binding global treaty. It needs to establish a minimum set of shared principles around transparency, human oversight and accountability that everyone can build from, regardless of where they are starting. Second, it must genuinely include people who are using AI in practice, not just governments and large technology companies. As Secretary-General Guterres has framed it, the question is whether humanity governs AI together or lets it govern us. That word "together" has to mean something. The most important governance questions I encounter day to day are not abstract, they are practical: when is it appropriate to use an AI tool on sensitive data? Who checks the output? What happens when it is wrong? Those questions come from real workplaces, and the Dialogue needs to create structured space for that ground-level experience to inform what gets agreed. Third, it must not leave less resourced nations behind. As CSIS analysis has highlighted, 118 countries remain absent from prominent international AI governance initiatives. A framework that only works for wealthier, more technically advanced nations does not solve the problem, it institutionalises a new version of it. Success means proving that very different countries, organisations and individuals can still build something coherent together. That is what this first Dialogue needs to demonstrate is possible.

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

Transparency, accountability and human oversight is my strongest priority. Resolution 79/325 calls for transparency, accountability and robust human oversight of AI systems in a manner that complies with international law and I see why this matters in practice. In my work, I regularly use AI tools to process and analyse data that informs senior decision-making. The question of who reviews the output, who is accountable when it is wrong, and how that process is documented is not abstract. It is something I navigate every week. International frameworks must address this operational reality, not just the principles behind it. Safe, secure and trustworthy AI connects directly to this. Working in a data-sensitive environment where information governance is critical, I have seen first-hand how trust in AI systems has to be earned through consistent standards and oversight not assumed. Without it, adoption stalls or, worse, proceeds without appropriate safeguards. Social, economic and ethical implications matter to me because my Level 7 apprenticeship has introduced me to how AI models can embed and amplify bias in ways that are not always visible. The people most affected by these implications are often the least represented in the rooms where governance decisions are made. That gap needs closing. AI capacity-building rounds out my selection. Building capacity in developing countries and addressing social, ethical and linguistic impacts is central to the resolution's equity agenda and as I work to build my own knowledge in this field, I understand how significant the gap between access and aspiration can be, even at an individual level.

Yes - two issues feel particularly important to me, and both come from my experience as a practitioner rather than a policy observer. The governance of agentic AI in everyday workplaces. The seven listed themes largely address AI as something organisations adopt and governments regulate. But the technology is moving faster than that framing allows. AI systems have moved beyond copilots and chat interfaces into agentic deployments tools that take actions, route decisions and operate continuously within workflows, often without direct human intervention at each step. am already seeing this in my own workplace with tools like Microsoft Copilot, which does not just answer questions but drafts, summarises and acts on data. The real risk is not model performance, it is the rapid proliferation of autonomous AI agents operating without governed identity, enforceable access controls or lifecycle governance. The current thematic areas do not yet name this directly. They should, because agentic AI changes the accountability question fundamentally. When an AI acts rather than advises, the governance frameworks designed for human decision-makers no longer map cleanly. AI literacy as a governance prerequisite. The listed themes focus on what governments and organisations should do. But governance ultimately depends on the people using these systems day to day having enough understanding to exercise meaningful oversight. Governance decisions are too often made without meaningful public consent, and policymakers and the public frequently lack the knowledge to provide effective oversight creating implementation gaps. I am currently building my own AI literacy through a postgraduate apprenticeship precisely because I recognised this gap in myself. That process should not rely on individual initiative. A cross-cutting theme on AI literacy not just for specialists but for practitioners across all sectors would strengthen every other theme on this list.

I am based in the United Kingdom and work in a complex, data-sensitive operational environment. My experience is at the practitioner level, I work with data governance frameworks daily and am studying AI and Data Analytics at postgraduate level. Both perspectives shape what I see as the most significant challenges and opportunities. The pace of adoption is outrunning the governance. The UK government's AI Opportunities Action Plan has set an ambitious direction, with the Prime Minister describing AI as a golden opportunity and commitments to upskill millions of workers. But as recent analysis has noted, only 26% of departments have integrated AI across their organisation, and the gap between ambition and execution remains wide. In high-stakes environments handling sensitive data, this gap is not just inefficient , it is a risk. Tools are being adopted before the governance frameworks to use them responsibly are in place. AI literacy at the practitioner level is critically underdeveloped. The Defence AI Strategy itself acknowledges this as a whole-of-sector challenge raising understanding at all levels, particularly improving AI literacy among policy, legal and commercial staff, and generating an informed user base with the knowledge and confidence to use new capabilities effectively. I see this every day. People are using AI tools without a clear understanding of what they should and should not put into them, who is responsible for checking outputs, or what the governance boundaries are. The opportunity, however, is real. The UK has strong institutional foundations: the Centre for Data Ethics and Innovation, the AI Safety Institute, and an active regulatory community. The question is whether governance frameworks can be made practical and accessible enough for people working at the operational level to actually apply them. That is the gap international dialogue needs to help close.

From where I sit , the most important role the Dialogue can play is to be the place where fragmentation stops. Right now, global AI governance is pulling in too many directions at once. The recent India AI Impact Summit made no reference to the Independent International Scientific Panel on AI or the forthcoming Geneva Dialogue, despite both being explicitly designed as the UN's central venues for evidence-based governance and multilateral coordination. When major summits do not even reference each other, the risk is not just inefficiency - it is that governance frameworks become incompatible, and the people expected to work within them are left with no coherent standard to follow. Secretary-General Guterres has put this plainly: "No country can see the full picture alone. We need shared understandings to build effective guardrails, unlock innovation for the common good, and foster cooperation." The Dialogue is uniquely positioned to provide that; not by replacing other initiatives, but by acting as the connective tissue between them. Three things would help it do that well. First, connect the dots between parallel processes, the India Summit, the Paris Declaration, Bletchley, and Geneva should be feeding the same framework, not competing with it. Second, use the UN's unique role to provide an umbrella where different approaches can be made interoperable, standards can be practical across different market realities, and the benefits of digital transformation remain open to all. Third, resist the pressure to produce only high-level declarations. Geneva 2026 must be the moment the international community moves beyond principles , turning evidence into commitments, and commitments into cooperation. As someone building their own understanding of AI governance from the ground up, I want a Dialogue that produces something I can actually point to and use.

There is no shortage of existing AI governance initiatives; the challenge is that they are not talking to each other effectively. The Dialogue's added value lies not in creating something new, but in doing what no other body has managed: connecting what already exists into something coherent. The foundations are genuinely strong. The OECD AI Principles established the first intergovernmental standard on AI and have influenced landmark regulatory efforts including the EU AI Act and the NIST AI Risk Management Framework. UNESCO's Recommendation on the Ethics of AI, endorsed by all 194 member states, provides a globally accepted normative framework grounded in human rights. The GPAI integrated partnership brings together 44 countries alongside governments, industry, academia and civil society to advance trustworthy AI. The Bletchley, Seoul and Paris summits have each moved the conversation forward. These should all feed into Geneva , not be replaced by it. The problem, as I noted in my previous answer, is fragmentation. Efforts remain fragmented, with gaps in how documentation and transparency requirements connect across the AI value chain - upstream developers, downstream deployers and end users are all working without shared expectations. From where I sit as a practitioner, this is a real daily challenge. The frameworks exist on paper but do not connect in practice. The Dialogue's unique added value is its universality. The UN-hosted Global Dialogue was set in motion specifically to broaden understanding beyond the groups already engaged , connecting the work of the OECD, GPAI, standards bodies and regional initiatives into a network that includes countries and voices currently absent from those rooms. Build on what exists. Fill the gaps between it. And make the results usable by the people actually working with AI day to day not just the organisations governing it from above.

The most important structural principle for the AI Dialogue is that different stakeholders have genuinely different things to contribute and the format needs to reflect that, rather than giving everyone the same kind of slot. Governments bring legal authority and the ability to commit to frameworks. Technology companies bring technical knowledge about how these systems actually work. But practitioners, the people using AI tools in real workplaces, making daily decisions about when and how to apply them , bring something different again: an understanding of where governance frameworks do and do not map onto operational reality. The credibility of the process rests on preventing dominance by a few states or corporations, and that means actively creating space for voices that do not arrive with delegations or PR teams behind them. I would recommend three structural changes. First, dedicated practitioner sessions , not just panels of experts talking about how AI is used, but structured input from people actually using it in public services, data-driven environments and operational settings. Second, an open written submission process that is actively promoted, not just technically available. This form is a good start, but most practitioners will never hear about it unless outreach goes beyond UN communication channels. Third, meaningful pre-session engagement so that submissions like this one actually feed into the agenda, rather than being received and filed. All parties : governments, the private sector, civil society, the technical and academic communities, and users must be involved in their respective roles. That is the right principle. The question is whether the format makes it genuinely possible, or just notionally inclusive. The Dialogue should be designed around that question from the start.

Several groups are consistently underrepresented in global AI governance discussions, and in my view three deserve particular attention. Practitioners at the operational level. The people implementing AI tools in workplaces analysts, coordinators, data professionals, public sector workers are almost entirely absent from governance conversations. Yet they are the ones navigating the real gaps between policy and practice every day. I am one of those people. I work with data governance frameworks, use AI tools professionally, and am studying AI and data ethics. I submitted this response because I happened to come across this form not because there was any structured effort to reach people like me. That gap needs to close. Women in AI governance. Women are underrepresented in the technology and AI dialogue, and this is not just a diversity concern, it is a quality-of-governance concern. The perspectives, risk tolerances and lived experiences that women bring to questions of accountability, bias and oversight are materially different, and frameworks designed without them will have blind spots. Organisations like Women in AI Governance are working to address this, and the Dialogue should actively partner with them. The Global South at civil society level, not just government level. The underrepresentation of civil society from the Global South in digital governance discussions excludes the local concerns of historically marginalised groups, leaving their rights unaccounted for in decision-making arenas affecting them. Government delegations from these nations do not always represent grassroots perspectives. Specific measures, including allocating funding and preparing budgets are necessary for marginalised groups to engage effectively in policy-making. Inclusion that requires people to self-fund their participation is not real inclusion. The test is simple: if the same voices dominate in Geneva that dominate everywhere else, the Dialogue has failed on this question.

The standard conference format : panels, keynotes, side events is not sufficient for a Dialogue that claims to be genuinely inclusive. I would suggest three formats that could meaningfully change who contributes and how. Practitioner input sessions with structured synthesis. Rather than asking practitioners to observe panel discussions, create dedicated sessions where operational-level input is gathered in advance through forms like this one, through structured interviews, or through facilitated regional workshops and then synthesised and presented back to delegates as evidence. AI agents should augment rather than replace direct human participation, serving as bridges to inclusion where traditional participation faces obstacles , and AI-assisted synthesis of large volumes of practitioner input could make this feasible at scale without reducing individual voices to statistics. Asynchronous and multilingual participation tracks. Not everyone who has something valuable to contribute can be in Geneva in July. A parallel online track genuinely interactive, not just a livestream with real-time translation and structured ways to respond to what is being discussed in the room would extend the Dialogue's reach significantly. The IGF Policy Network on AI has demonstrated that ensuring diverse perspectives, especially from underrepresented countries or regions, are heard in global AI dialogues requires deliberate structural design, not just good intentions. Open challenge or case study submissions. Invite practitioners, civil society organisations and individuals to submit real-world governance challenges they have encountered anonymised where necessary and use these as the working material for breakout discussions. This grounds abstract governance debates in concrete problems and ensures the Dialogue is solving for reality, not just principles. The format of a conversation shapes who feels able to speak in it. Geneva 2026 should be designed for the people who are not usually in the room.

Several existing approaches stand out to me as genuinely useful not just as policy documents, but as frameworks that map onto how AI is actually being used in practice. The NIST AI Risk Management Framework is one of the most practically useful tools available. Its four functions : Govern, Map, Measure, Manage give organisations a structured, adaptable way to think about AI risk that works across different sectors and scales. I have studied it as part of my postgraduate apprenticeship and found it to be one of the few frameworks that translates well from principle to practice. The Dialogue should actively promote its adoption and support nations in building capacity to apply it. The EU AI Act's risk-tiered classification system , categorising AI systems as unacceptable, high, limited or minimal risk offers a concrete, enforceable structure that other jurisdictions can learn from. Its phased implementation approach is also instructive: rather than demanding full compliance overnight, it gives organisations time to build the infrastructure governance requires. In the UK, the Centre for Data Ethics and Innovation's AI Assurance Framework establishes core principles around lawful purpose, technical robustness, fairness, transparency and contestability and provides practical guidance to help organisations deploy trustworthy AI systems rather than just articulating what trustworthy means in theory. That distinction between stating principles and helping people implement them is where most governance frameworks fall short. What these approaches share is that they move governance from aspiration to operation. The Dialogue should not produce more principles documents. It should identify what is already working, document why it works, and support nations and organisations , particularly those with less capacity to adopt and adapt these approaches for their own contexts. That is what effective international cooperation on AI governance actually looks like in practice.