America can’t win the AI race if the rollout lacks public support

The writer is a fellow at the Johns Hopkins School of Advanced International Studies

America is finally having a serious national reckoning over who gets to set the rules for a transformative technology. The Meta legal agreement, reached last Wednesday, will see the company pay up to $18bn to 48 US states and Washington DC for harm caused to children.

America should not make the same mistake with AI that it made with social media. Fundamental questions about Big Tech’s responsibilities cannot only be adjudicated after its platforms have already changed the lives of an entire generation. The country should learn from this experience before this new technology becomes even more deeply embedded in work, education, government and everyday life.

This points to a central question for the AI age. Who gets to decide how AI will shape America’s future? So far, the answer is tech companies and politicians. The public is left a bystander, even though anxieties about AI cross Democratic and Republican divides. Opinion polls reflect increasing scepticism and distrust around the technology. Protests against data centres are spreading. Tech CEOs have been booed and had their homes attacked.

This is not a promising start to governing a technology predicted to affect nearly every aspect of life. America needs a process equal to the scale of this revolutionary moment. It needs a national dialogue on AI, one that can guide the country through a time of dramatic change.

Building this future is not merely a technical exercise but a social and moral one. Costa Rica offers a useful model, having been the first nation to engage in a democratic conversation on AI. In 2025, it convened a National Dialogue on AI, bringing together civil society, government, business and academia. The dialogue took an ethics-based approach which emphasised balancing national competitiveness with human dignity, oversight and transparency. Iceland is following suit and will hold a national dialogue on AI this autumn.

In 2026, the UN hosted the first Global Dialogue on AI Governance, which brought together various government, private sector and civil society actors to discuss issues of access, safety, accountability and human oversight. But a global conversation is not a substitute for a national one.

The White House and Congress should launch a bipartisan, one-year National Dialogue on AI. The process requires the imprimatur of government. However, it should be run by an independent commission with representation from politics, tech, business and labour, supported by scientific, educational and civic institutions.

The commission should begin with an agenda-setting process which establishes a transparent mandate and ground rules. The consultations should engage with the full diversity of American society, combining outreach with channels for feeding public input into the process.

The dialogue needs to go further than the conversations we have seen in Costa Rica and result in a national AI roadmap designed to translate consensus into action. This should feed into legislation, with clear outcomes and deadlines as well as implementation and monitoring mechanisms.

Some might say that AI is too complex for public deliberation or that it is moving too fast and neither the market nor China will wait for town hall debates. Others will warn that a national dialogue will turn into a partisan show, or become an exercise in legitimising decisions already made by power brokers in Silicon Valley and Washington.

These are serious concerns. But the public does not need to understand every detail of how AI works in order to shape the course of the technology and the future of American society.

A well-designed dialogue would in fact benefit not just citizens but also Silicon Valley and the US government. It would provide the tech industry with legitimacy and predictability in place of the ad hoc regulations increasingly appearing at the state and federal levels. In 2025, state legislatures passed 196 laws related to AI. In 2026, 137 AI-related bills have already been enacted.

For the government, a dialogue would provide a public mandate for developing a national AI agenda. AI governance based on democratic legitimacy would also give the US an advantage that the Chinese system cannot offer.

America cannot win the AI race if Americans come to believe the AI revolution is something being done to them rather than built with them. Public legitimacy isn’t an impediment to American competitiveness; it is ultimately necessary for it.