AI: What Could We Accomplish Together?

Much of the international conversation about artificial intelligence begins with what might go wrong.

There are good reasons for that. AI introduces consequential questions about safety, security, accountability, economic disruption, and the concentration of power. Some risks may cross national borders, and no country can assume its own rules will be sufficient.

But there's another question worth asking.

What might become possible if nations approached AI not only as a potential threat to manage, but as a growth opportunity to pursue together?

Consider health care. Scientific discovery. Education. Food security. Climate research. The ability to anticipate natural disasters, improve public services, and make useful knowledge available to people who have historically had limited access to it.

None of these outcomes are guaranteed. Better technology does not automatically produce better lives. But the possibilities are substantial enough to deserve serious international attention.

Cooperation is not a new idea.

Nations have repeatedly found ways to cooperate when the consequences of acting alone became too great—or the potential benefits of acting together became too compelling.

The World Health Organization provides a forum for international public-health cooperation. The World Bank supports development across national boundaries. The Paris Agreement established a framework for collective action on climate change.

Even international sporting institutions show our willingness to set common rules and pursue shared endeavors despite profound differences.

These examples are imperfect. So are humans. So is AI. Their mandates, authority, effectiveness, and political circumstances differ considerably.

We are not suggesting that any one of them provides a ready-made model for AI.

We are suggesting that the underlying proposition is familiar: nations can compete, disagree, and still recognize interests that warrant cooperation.

What would we cooperate to accomplish?

The discussion should begin with outcomes, not organizational charts.

Could international research partnerships accelerate progress against diseases that affect millions?

Could shared scientific resources help researchers in less wealthy countries contribute to discoveries rather than consume technologies developed elsewhere?

Could AI improve access to education in underserved communities, including in languages and settings that commercial markets have little incentive to prioritize?

Could better coordination make it easier to anticipate and respond to disasters whose consequences do not respect borders?

These are propositions worth investigating, not promises to accept on faith.

Each would require evidence about what AI can actually accomplish, the conditions necessary for success, the costs and risks involved, and who would benefit.

And each raises questions of authority, accountability, and access.

Who gets a seat at the table?

The countries and companies developing the most powerful AI systems will inevitably play an important role.

But technological capability should not be the only qualification for participation.

Countries with fewer resources may have different priorities, different institutional constraints, and different definitions of what useful progress looks like. Their populations will nevertheless experience the consequences of decisions made elsewhere.

An international effort that concentrates decision-making among a handful of governments and technology companies would risk reproducing the very inequalities that cooperation might help address.

Participation must be meaningful. So must access to the benefits.

That does not mean every country will contribute identical resources or exercise identical authority. It means that the people affected by these decisions deserve representation in how their consequences are understood and governed.

The obstacles are real.

International cooperation is difficult even when the objective is widely shared.

AI adds significant complications. Nations see strategic and economic advantages in developing superior capabilities. Companies face commercial incentives to move quickly. Governments have legitimate national-security concerns, including the possibility that capabilities shared for beneficial purposes could be misused.

Trust cannot simply be declared into existence.

Nor should voluntary commitments be mistaken for adequate oversight in every circumstance.

The strength of the evidence required, the safeguards imposed, and the independence of oversight should reflect the consequences of a particular use of AI and the authority being delegated to it.

We should be prepared to distinguish between areas where voluntary cooperation may be sufficient and those where enforceable obligations are necessary.

We should also acknowledge what we do not yet know.

We do not know the ideal institutional architecture for international AI cooperation. We do not know which opportunities will justify the investment or which arrangements will survive competitive pressures.

Those uncertainties are reasons to investigate, experiment, and learn—not reasons to abandon the question.

An opportunity worth organizing around

Safety and opportunity are not competing objectives.

The same seriousness we bring to preventing harm should inform our efforts to produce demonstrable public benefit.

That means defining intended outcomes, testing the evidence, identifying material uncertainties, assigning responsibility, measuring results, and revisiting decisions when circumstances change.

It also means resisting two temptations: assuming AI will solve problems simply because it is powerful, and dismissing potential benefits because the technology introduces unfamiliar risks.

Neither position substitutes for evidence.

International cooperation on AI should be ambitious enough to pursue meaningful progress and disciplined enough to determine whether that progress is actually occurring.

The world has consequential problems that no single nation can solve easily. We also have a rapidly developing technology that may help us address some of them.

Surely that combination deserves more than a conversation about how to keep the worst from happening.

We don't have all the answers; we know that. But we do have questions, and that's good. Knowing we have to do something is even better.

If the world's nations can come together to confront shared threats, they can also come together to pursue a shared opportunity. What are you prepared to do?