Artificial intelligence (AI) is often discussed in terms of algorithms, computing power, datasets and innovation. But global conversations about AI governance show that AI is also about infrastructure, participation, human rights and power: who is connected, whose knowledge is represented and who gets to shape our digital future.
From 6 to 7 July 2026, I participated in the first ever United Nations Global Dialogue on Artificial Intelligence Governance in Geneva, Switzerland. The gathering brought together governments, UN agencies, researchers, companies, technical experts and civil society organisations to discuss AI governance, digital infrastructure and inclusive digital transformation. Across the discussions, one lesson stood out: AI cannot contribute meaningfully to development unless the foundations required to support it are in place.
For communities in the Global South, conversations about AI must begin with meaningful and affordable connectivity, reliable energy, digital skills, locally relevant data and institutions capable of protecting people’s rights. Communities must also be recognised not simply as users of technologies designed elsewhere, but as people with the knowledge, experience and right to shape how technologies are built and governed.
Who has the connectivity to participate in AI?
During the Human Rights Cluster, I spoke about the relationship between inclusive AI, connectivity and community power. I introduced the work of the Local Networks initiative, a collaborative initiative of the Association for Progressive Communications and Rhizomatica that works alongside communities to strengthen community-centred connectivity. Discussions about inclusive AI frequently focus on bias, datasets, governance and representation. These are essential issues. But we must begin with another question:
Before communities can contribute data, develop locally relevant applications, influence policy or benefit from innovation, they need connectivity that is affordable, meaningful, trusted and responsive to their realities.
Community-centred connectivity is about more than providing internet access. It involves communities participating in the ownership, management and governance of the infrastructure through which they connect. It can strengthen resilience, support local languages and knowledge, and create the conditions for meaningful digital participation.
Equity, rights and power
Connectivity is therefore not only a technical issue. It is a question of equity, rights and power.
Across the Global South, many women, rural communities, Indigenous peoples and speakers of underrepresented languages remain excluded from digital spaces. When communities are excluded from connectivity, they are also excluded from shaping AI. Technologies presented as serving everyone may consequently reflect only a narrow range of languages, experiences, knowledge systems and priorities.
Community networks, grassroots organisations and local innovators are not simply beneficiaries of AI development; they are essential partners in building and governing it.
Communities are the first mile
Several discussions in Geneva raised concerns about the concentration of AI capabilities among a small number of companies and countries. Countries may gain access to AI systems while remaining dependent on technologies they cannot inspect, adapt or govern. Communities may be consulted only after a system’s purpose, design and rules have already been determined. Access under these conditions should not be confused with inclusion. Communities must be involved from the beginning: when problems are identified, research is designed, data is collected, and safeguards and accountability mechanisms are developed.
Describing communities as the “last mile” suggests that they are merely the final destination for infrastructure and technologies created elsewhere. A community-centred approach understands them as the first mile: the starting point for identifying priorities, languages, knowledge and needs.
Locally governed infrastructure can also support digital sovereignty by giving communities greater influence over decision making, stronger control over their data and more opportunities to develop applications that respond to their social, economic and cultural realities.
Access without agency is not inclusion
The UN dialogue also demonstrated that participation in global AI governance cannot be measured only by counting how many civil society organisations, women or representatives from developing countries are invited. We must also ask who can physically enter decision-making spaces, whose knowledge is treated as credible and who has the power to influence outcomes. Visa barriers, limited travel support, language differences and inaccessible policy processes prevent many people from the Global South from participating meaningfully, even when international processes claim to be inclusive.
There is an important difference between access and agency. Access means being invited to a meeting, using technology or participating in a consultation. Agency means being able to influence the agenda, challenge assumptions, propose alternatives and hold decision makers accountable.
Inclusive governance therefore requires more than representation: it requires changes in how power is distributed.
Hybrid participation, decentralised consultations, travel support and sustained engagement with grassroots organisations can make governance spaces more accessible, but they must be accompanied by a genuine willingness to share decision-making power.
A feminist approach asks who holds power
Gender and human rights activists were present throughout the Geneva discussions, although they were not always treated as central to AI governance. Women remain underrepresented in AI research, technical development, leadership and policy making. Women human rights defenders, journalists, activists and other historically marginalised groups may also face particular forms of technology-enabled discrimination, surveillance and violence. These harms cannot be addressed through general promises of inclusion.
A feminist approach asks deeper questions: Who defines the purpose of an AI system? Who owns the data used to build it? Whose knowledge is considered valuable? Who receives the benefits? Who carries the risks? Who can obtain remedy when harm occurs?
The Feminist Guiding Principles and Commitments for Global AI Governance, developed by the Gender in Digital Coalition, provides a framework for translating these questions into action. They call for safety and dignity by design, precaution before deployment, feminist data governance and sovereignty, human rights compliance, meaningful diversity throughout the AI lifecycle, and investment in feminist and community-led AI ecosystems.
These principles are directly relevant to community-centred connectivity. Feminist data sovereignty cannot be realised when communities lack control over the infrastructure through which their information is collected, stored and transferred. Community networks can help translate feminist principles into practice through locally governed infrastructure, community-led data stewardship, digital literacy and local-language innovation. But communities need the authority and resources to lead this work. Participation cannot be reduced to symbolic consultation or depend on unpaid community labour.
Connectivity must create public value
Discussions in Geneva also connected AI readiness with the practical delivery of public and community services. AI may support learning, healthcare, agriculture, assistive technologies and access to knowledge. However, these benefits are not automatic. Outcomes depend on the quality of local data, language and cultural relevance, the capacity of public institutions, and safeguards against bias, exclusion and unreliable outputs. AI should complement, rather than replace, frontline workers, local knowledge and accountable institutions.
Government interventions highlighted the importance of connectivity, digital public infrastructure, local datasets, skills, institutional capacity and human oversight. They also demonstrated that local and regional authorities should be treated as governance actors, not merely as implementers of technologies procured elsewhere.
Local services should therefore become a public-value test for AI and connectivity investment. The central question is not simply whether a community can access an AI tool. It is whether the wider service ecosystem can improve outcomes in ways that are equitable, reliable and locally governed.
This means beginning with community-defined service needs rather than selecting a technology first. Communities, schools, clinics, cooperatives, local authorities and other frontline institutions should participate as co-designers and governance partners. Initiatives should prioritise low-bandwidth, offline-capable and local-language systems. Where AI affects access to rights or essential services, there must be human oversight, transparent procurement, independent testing, accessible grievance mechanisms and clear responsibility when harm occurs. Success must also be measured beyond connection numbers. Relevant indicators include affordability, service quality, inclusion, trust, accountability, error rates and strengthened local capacity.
Language, data and community knowledge
The dialogue happened alongside the World Summit on the Information Society Forum. The forum provided practical examples of how countries are linking broadband, digital skills and local AI development. A presentation on Nigeria’s Project BRIDGE demonstrated that international connectivity reaching a country does not necessarily mean that schools, health facilities, public institutions and rural communities have reliable broadband. The presentation also highlighted investment in datasets for Hausa, Igbo, Yoruba and Nigerian-accented English. This illustrates why infrastructure investment must be connected to local-language development.
When digital systems do not understand the languages people use, entire communities may be excluded from services and opportunities. But creating language datasets is not merely a technical exercise. It raises questions about consent, ownership, representation and who benefits from community knowledge.
Communities should participate in decisions about how their languages, voices and data are collected and used. They should not simply provide raw material for systems owned elsewhere. Reliable energy must also be part of this discussion. Connectivity, AI services, digital public infrastructure and cloud systems all depend on electricity. Investment in digital infrastructure must therefore be connected to sustainable energy, environmental accountability and community resilience.
Building AI from the community upwards
The lessons from Geneva reinforce several priorities for the wider community-centred connectivity movement. Community networks should be recognised as foundational infrastructure for equitable AI participation. Grassroots communities must be involved when agendas, research and policies are being developed, not only during implementation. Investment in connectivity should be linked to local services, digital skills, local-language technologies, feminist data governance, sustainable energy and community-led innovation.
Most importantly, the future of AI cannot be determined only in corporate laboratories, data centres and international conference halls. It must also be shaped by rural and Indigenous communities building and governing their own networks; by women developing technologies that respond to local realities; by speakers of African and other underrepresented languages; and by organisations working to advance human rights and digital justice.
As I concluded in my intervention to the Human Rights Cluster:
There is no inclusive AI without inclusive connectivity. And there is no inclusive connectivity without communities at the centre.
For AI to serve everyone, the communities most often excluded must be able to shape the digital future from its foundations.
Equitable AI begins long before an algorithm is deployed. It begins with meaningful connectivity, reliable energy and locally relevant infrastructure. It begins with services that respond to community-defined needs. It begins with gender justice, human rights and accountability.
Above all, it begins with communities.
Rebecca Ryakitimbo is a feminist technologist and researcher working at the intersection of AI, gender justice and digital equity. She supports feminist tech spaces such as the African Women School of AI, and curates the Gendering AI conference. As part of the Local Networks (LocNet) initiative, she supports community-centred connectivity initiatives by facilitating communities of practice and researching community-centred connectivity and local services for equitable, locally led digital ecosystems.