The Future of AI Is Everyone’s Responsibility

In Onna Village, on Okinawa’s western coast, volunteers with Team Chura Coral restore damaged reefs by planting coral seedlings. At the University of the Ryukyus, researchers study water quality and help communities understand the ecosystems they depend on.

These initiatives rarely attract attention. Yet they illustrate a fundamental truth: global challenges are experienced locally, and effective solutions often begin with those closest to the problem.

During my recent visit to Okinawa to lecture on artificial intelligence (AI) and the Sustainable Development Goals (SDGs), I was reminded that communities experiencing environmental degradation, water insecurity and technological change hold essential knowledge for finding solutions.

This raises a question: who gets to shape the future of AI?

Today, a handful of companies and governments hold much of the influence over technological development and regulation. Silicon Valley shapes what AI can do, while Brussels increasingly determines how it is regulated. Yet the consequences are felt worldwide.

A technology that transforms humanity cannot be governed by only a fraction of humanity.

Promise and responsibility

AI offers extraordinary opportunities for sustainable development. It can help farmers identify pests, support medical diagnosis, personalize education and improve disaster forecasting. For island communities facing typhoons, rising seas and environmental pressures, these capabilities can strengthen resilience.

But technological capability does not automatically produce social progress. Without appropriate governance, AI can reproduce inequalities, concentrate power and generate new environmental costs.

Three challenges demand attention.

The first is accountability. Many AI systems operate through processes that are difficult to explain. When an algorithm wrongly denies employment, recommends inappropriate treatment or produces discriminatory outcomes, who is responsible?

An algorithm’s complexity cannot excuse the erosion of accountability. Effective governance requires human oversight, transparent institutional responsibilities, mechanisms to contest decisions, and remedies when systems fail.

Data governance raises similar concerns. Personal information has become an economic resource, frequently collected and monetized without meaningful individual control. The central question is who determines how data is collected, processed and used, and who benefits.

The second challenge is environmental sustainability.

AI is promoted as a climate solution, yet its infrastructure consumes electricity, water, minerals, and computing hardware. The International Energy Agency projects that global data centre electricity consumption could reach approximately 945 terawatt-hours annually by 2030, with AI driving substantial growth.

Data centre cooling, semiconductor manufacturing, mineral extraction and electronic waste create additional environmental pressures. These costs are unequally distributed: communities benefiting least from AI may bear significant burdens from its infrastructure.

Responsible governance must therefore assess AI’s entire lifecycle, from resource extraction and model training to deployment and disposal.

We cannot call AI sustainable merely because some applications benefit the environment. Its infrastructure must also be sustainable.

The third challenge is inequality.

The digital divide increasingly concerns power, not simply connectivity. UNICEF and the International Telecommunication Union report that approximately two-thirds of school-age children worldwide lack internet access at home.

Yet connectivity alone is insufficient. Countries may remain dependent on foreign computing infrastructure, proprietary models, and data ecosystems controlled externally.

This creates a new challenge of AI sovereignty: ensuring societies possess sufficient technical, institutional and regulatory capacity to make meaningful choices about technologies shaping their futures.

AI models trained predominantly on data from wealthy societies may inadequately represent other languages, cultures, environments and institutions. Agricultural systems may misunderstand tropical conditions; healthcare models may perform unevenly across populations; language technologies may marginalize indigenous knowledge.

These are questions of representation and power.

The Global South must not merely consume AI but also develop its infrastructure, produce its knowledge and govern its consequences.

Governance must be global and inclusive

Four principles should guide international AI governance.

First, inclusion: governments, researchers, businesses, civil society, workers and local communities must participate meaningfully in setting priorities and evaluating risks.

Second, adaptability: rapidly evolving technologies require continuous assessment and regulatory learning, not frameworks that remain unchanged for decades.

Third, evidence: independent scientific evaluation must guide policy, replacing both technological hype and exaggerated fears.

Fourth, international coordination: common safeguards should protect human rights and accountability while respecting different development priorities and institutional circumstances.

The OECD AI Principles and the European Union AI Act are important contributions, but fragmented regulation risks inconsistency and regulatory arbitrage.

The objective should be shared principles, interoperable safeguards, and meaningful local agency, not uniformity at all costs.

The United Nations has begun laying the groundwork for this cooperation. The Global Digital Compact, adopted in 2024, advances international digital cooperation. The Global Dialogue on AI Governance and the Independent International Scientific Panel on AI offer mechanisms for more inclusive discussion and shared scientific understanding.

At the United Nations University, we contribute through research, education, and policy engagement. We are developing our new Institute for Artificial Intelligence in Bologna to connect expertise from the Global North and South. Our free course, Responsible AI in a Global Context, developed with Microsoft, broadens access to essential AI governance knowledge.

Such initiatives matter because meaningful participation requires more than representation. It requires the capacity to understand, evaluate and influence technology.

Why humans matter

The students I met in Okinawa belong to a generation that will live alongside AI throughout their lives.

My message was simple: do not surrender your judgment to an algorithm.

AI can predict coral reef deterioration, but it cannot independently determine what a reef means to a fishing community or what obligations one generation owes another.

It can optimize resource allocation, but it cannot determine what constitutes a just distribution.

These decisions require human judgment, cultural understanding and democratic accountability.

Coral restoration volunteers and water researchers in Okinawa show that valuable knowledge exists far beyond major technological centers. Their experience must inform how we design and govern intelligent systems.

The future of AI is not predetermined. It will reflect decisions about who controls infrastructure, whose knowledge is represented, how environmental costs are distributed and who establishes the rules.

Silicon Valley will remain a center of innovation. Brussels will remain influential in regulation. Both have important contributions to make.

But the future must also be shaped in African universities, Asian laboratories, Pacific island communities and institutions throughout the world.

Our task is to ensure AI becomes not only more powerful but more accountable; not only more efficient but more sustainable; and not only more intelligent but more just.

Humanity as a whole must write the future of AI.