Sustainable AI for Sustainable Societies: The Missing Dimension in Global AI Governance

By Luís Neves, Chief Executive Officer, GeSI 

The AI for Good Summit has become an important barometer for how the world is thinking about artificial intelligence. It is no longer simply a forum about technology, applications, or innovation. Increasingly, it reflects a deeper question: who will write the rules for the next industrial revolution? 

Behind the common language of “AI for humanity” sit different geopolitical, economic, and strategic interests. Governments are concerned with sovereignty, security, and competitiveness. Industry is focused on innovation speed, investment certainty, and global markets. The United Nations brings attention to inclusion, human rights, and the risk of fragmentation. Scientists raise questions about capability risks, safety, and alignment. Civil society focuses on fairness, accountability, and access. Developing countries are asking how to avoid a new digital divide. 

Everyone agrees that AI is transformational. There is far less agreement on who should shape its future, how responsibility should be allocated, and how quickly governance can evolve. 

From “Can AI Help?” to “How Do We Govern AI Responsibly?” 

Five years ago, much of the AI for Good conversation centred on use cases. The focus was on how artificial intelligence could improve healthcare, education, agriculture, disaster response, climate action, and public services. 

Today, as AI capabilities continue to advance, the conversation has naturally evolved. Alongside innovation, there is growing attention to governance, trust, transparency, and the frameworks needed to ensure AI delivers positive outcomes for society. 

AI is increasingly recognised as foundational digital infrastructure, with the potential to reshape economies, industries, and public services. As a result, discussions now extend beyond technology itself to include questions around standards, interoperability, accountability, resilience, and international cooperation. 

This evolution reflects an important reality: the future of AI will depend not only on technological progress, but also on our collective ability to develop, deploy, and govern it responsibly. 

The AI Stack Is the Real Strategic Battleground 

The public often sees AI through applications, chatbots and models. But the real competition lies across the full AI stack. 

That stack includes chips, cloud infrastructure, data, energy, talent, foundation models, standards, and regulation. Countries and companies that dominate several of these layers will have far more influence over the future than those that only build applications on top. 

This explains why export controls, semiconductor policies, sovereign compute, energy investments, and data governance have become strategic priorities. The true strategic advantage lies not only in model capability, but in the ability to build a resilient, trusted, and scalable innovation ecosystem. 

Broad Consensus on Principles, Limited Consensus on Implementation 

There is remarkable agreement on the broad principles that should guide AI. 

Most actors now support human-centric AI, transparency, safety, accountability, privacy, inclusion, sustainability, and international cooperation. These principles are important. They show that a common language is emerging. 

But principles are not implementation. 

The taught questions remain unresolved. Who certifies AI systems? Who audits foundation models? What constitutes unacceptable risk? How should open-source frontier models be managed? Which authority applies when AI systems operate across borders? Who owns the data? Who pays for safety? How do we ensure that systems with global societal impact remain accountable when their development is concentrated among few actors? 

These are not abstract questions. They will shape trillions of dollars in future economic value and determine whether AI develops in a way that strengthens societies or deepens fragmentation. 

Global Governance, Not Global Government 

A single global AI regulator is unlikely to emerge. It would be unrealistic to expect one institution to govern all aspects of AI across every region, sector and use case. 

But global governance is different from global government. 

The world already operates through layered governance in other domains, including civil aviation, telecommunications, maritime law, financial regulation, and nuclear safeguards. No single authority controls these systems globally. Instead, they rely on shared standards, treaties, interoperability, sector-specific institutions, and trusted cooperation. 

AI may evolve in a similar direction. 

We are likely to see global principles, regional regulations, sector-specific rules, and technical standards operating together. Healthcare, finance, transport, education, energy, defence, and public services will each require different forms of oversight. International standards bodies will play a role in testing, assurance, and interoperability. Regional frameworks will reflect local priorities and legal systems. 

The challenge is not to create one universal regulator. The challenge is to create enough coherence to avoid fragmentation while preserving innovation, trust, and accountability. 

The Missing Dimension: AI’s Environmental Footprint 

Yet while policymakers grapple with data sovereignty, AI safety, model accountability, and regulatory fragmentation, one critical issue remains underrepresented in global AI governance discussions: the environmental footprint of AI itself. 

AI is increasingly presented as a climate solution. It can help optimize electricity grids, accelerate renewable energy integration, improve weather and disaster prediction, increase industrial efficiency, support precision agriculture, and reduce emissions across sectors. 

While this narrative is both true and important, it risks eclipsing the other side of the equation: AI’s own physical and environmental footprint linked with the rapidly expanding infrastructure, data centres, high-density computing, electricity generation, cooling systems, transmission capacity, critical minerals, hardware manufacturing, and end-of-life management. 

The discussion seems having been shifted from “AI consumes too much energy” to “we need to build enough clean energy for AI.” That is a necessary response, but it sidesteps the deeper question: how should societies govern the energy, water and material demands of AI as it scales? 

If AI becomes one of the world’s most strategic infrastructures, then its own resource demands must be governed with the same seriousness as its social, ethical, and economic impacts. 

Several questions deserve much greater attention, such as:. 

How much energy should society be willing to dedicate to AI? Who gets access to scarce clean electricity as AI data centres compete with the electrification of transport, housing, and industry? How should AI’s lifecycle footprint be measured across training, inference, hardware, water use, and supply chains? Can AI itself become a rebound effect, where efficiency gains are offset by much higher overall consumption of compute and electricity? 

These are not only technical questions. They are governance questions. They are also fundamental sustainability questions. 

From AI for Sustainability to Sustainable AI 

This is where GeSI sees an important opportunity. 

The current global conversation often focuses on AI for sustainability: how AI can help solve climate and sustainability challenges. That agenda is essential. AI can support mitigation, adaptation, resilience, efficiency, and better decision-making. 

But we must also ask a second question: how can AI itself be developed, deployed, and governed within planetary boundaries and in ways that are socially equitable? 

While related, “AI for Sustainability” and “Sustainable AI” are fundamentally different. The former asks how AI can help address climate and sustainability challenges and is receiving growing attention. The latter asks how the AI ecosystem itself can remain within planetary boundaries — a question that remains far less developed as AI scales globally… and fast. 

For GeSI, this distinction matters. The sustainability challenge is no longer only how AI can help save the planet. It is also how we ensure that the infrastructure powering AI remains compatible with the planet’s ecological limits. 

The future depends not only on sustainable AI applications, but on a sustainable AI ecosystem. 

The Role of GeSI 

Organizations such as GeSI have a significant role to play in this emerging governance landscape. 

Governments negotiate. Companies innovate. The UN convenes. Standard bodies define technical pathways. Civil society raises questions of fairness, rights, and accountability. 

GeSI can help bridge these communities by translating broad principles into practical implementation. This means bringing together industry, policymakers, international organizations, and experts to develop measurable frameworks, evidence-based guidance and collaborative approaches that embed sustainability and human-centric values into AI deployment. 

This intermediary role is increasingly valuable because many of the most difficult AI governance questions are not purely legal or technical. They require trusted collaboration across sectors. 

They also require measurement. 

If AI is to be credible as a force for sustainable development, its benefits and impacts must be assessed with transparency. We need better ways to understand not only what AI enables across the economy, but also what the AI infrastructure consumes, how it is powered, how it is cooled, how hardware is sourced, and how social and environmental impacts are distributed. 

This is where GeSI’s long-standing work on digital sustainability, Digital with Purpose, sustainable digital infrastructure, data centres, supply chains, and impact measurement can contribute to the next phase of the AI conversation. 

A Trusted System of Innovation 

The AI for Good Summit shows that the world is converging on a shared destination, but not yet on a shared route. 

AI will transform society. That is no longer in question. The real question is whether governance can evolve quickly enough to ensure that this transformation remains aligned with human values, supports sustainable development, preserves innovation, and avoids geopolitical fragmentation. 

The race for AI leadership is no longer about building the smartest models. It is about building the most trusted system of innovation. 

The nations, companies and institutions that succeed will not simply lead in capability. They will lead in earned legitimacy. 

And that legitimacy will increasingly depend on whether AI can serve humanity without exceeding the ecological and social boundaries on which humanity depends.