AI governance frameworks built solely on expert principles are missing a critical ingredient: the public values that communities actually care about. Rather than debating abstract concepts like "fairness" and "transparency," policymakers and technologists should ground AI regulation in the deeply held beliefs that already shape laws, constitutions, and court decisions across society.

What Are Public Values, and Why Do They Matter for AI?

Public values are the shared beliefs that form the foundation of how a political community operates. According to policy scholar Barry Bozeman, they represent "those values providing normative consensus about the rights, benefits, and prerogatives to which all citizens should be entitled; the obligations of citizens to society; and the principles on which governments and policies should be based".

In a survey of more than 2,000 U.S. citizens, Bozeman found remarkable consensus on core values. Over 90 percent of respondents supported substantive principles including freedom of speech, physical liberty, equal access to civil rights, freedom of religion, gender equality, and physical safety and security. Over 80 percent supported protection of minority interests, access to health care, economic opportunity, and privacy.

These values already exist in the fabric of society. They are written into constitutions, major legislation, and court decisions; they appear in cultural touchstones and political speeches. The challenge is connecting them to emerging technology policy in ways that feel authentic to communities, not imposed from above by experts.

Why Do Abstract AI Ethics Principles Fall Short?

The AI field has converged around a set of widely endorsed ethical principles: privacy, accountability, safety and security, transparency and explainability, fairness and nondiscrimination, human control of technology, professional responsibility, and promotion of human values. On the surface, these sound uncontroversial. But the moment you ask what they mean in practice, consensus collapses.

Everyone supports "fairness" in the abstract, but people disagree sharply about what is fair in practice. Look at decades of debate over affirmative action in hiring and college admissions, or reparations for descendants of enslaved people. Similarly, "promoting human values" means almost nothing on its face. Liberty and equality are both broadly held human values, yet pursuing one can come at the expense of the other.

The real problem is that different groups benefit and suffer differently from AI systems. What benefits managers may harm workers. What helps one firm may hurt competitors. Urban and rural communities face different risks. The wealthy and poor experience AI's consequences in fundamentally different ways. In these contexts, agreement on AI ethics can only be reached on the "tamest of terms," according to one analysis of the Center for AI Safety's "Statement on AI Risk," which achieved consensus on a single sentence: "Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war".

How Can Communities Actually Shape AI Governance?

Rather than relying on opinion polling, which captures only a snapshot of public sentiment on a rapidly evolving issue, experts recommend deliberative methods that give citizens space to learn about technologies, discuss them with peers, and articulate their own priorities over time. These approaches have emerged from decades of research in technology assessment, responsible innovation, and anticipatory governance.

Effective methods for uncovering public values include:

• Public Forums: Structured conversations where citizens can engage directly with technology experts and policymakers about emerging AI systems and their potential impacts.
• Scenario Planning: Exercises that help communities envision different futures shaped by AI and discuss which outcomes align with their values.
• Integrating Humanistic Researchers: Embedding social scientists, ethicists, and humanities scholars directly into technical research and development processes to surface value questions early.
• Deliberative Engagement: Extended discussions that give citizens time to learn, reflect, and develop considered positions less prone to rapid shifts in opinion.

The key insight is that deliberative methods help citizens and researchers discover priorities together and articulate how today's technological issues connect to more durable conceptions of the good society and the good life.

What Role Does Existing Law Play in AI Governance?

One critical misconception is that AI regulation requires entirely new legislation. In fact, existing legal frameworks already apply in many cases. Anti-discrimination law, consumer protection, and copyright law are all being tested in relation to AI. Litigation involving data scraping and fair use is expanding, and cases involving harms linked to AI interactions are beginning to reach courts.

Congress has already acted in at least one area, passing the Take It Down Act to address nonconsensual intimate imagery, both real and AI-generated. States are increasingly seeking to regulate AI given the federal government's deregulatory approach.

However, law alone cannot do all the work many people expect. For rules to be implemented effectively, the infrastructure around them must exist: testing frameworks, evaluation systems, shared standards, and assurance mechanisms that allow institutions to determine whether rules are being followed. This structure also gives policymakers in different regions flexibility to decide whether such rules should be voluntary or mandatory, and helps foster regulatory interoperability when leaders in different jurisdictions disagree.

"Sometimes having a law when nobody knows what to do to implement it doesn't help as much," explained Lee Tiedrich, an inaugural fellow at the Artificial Intelligence Interdisciplinary Institute at the University of Maryland and visiting professor at the College of Information.

Lee Tiedrich, Visiting Professor of the Practice at University of Maryland College of Information

Tiedrich drew an analogy to the FDA. Public trust in prescription drugs does not depend on every individual understanding the science behind them. It depends on the existence of a system for testing, review, and oversight. For AI, the equivalent governance infrastructure is still being built.

Can Polarized Societies Find Common Ground on AI?

Even though Americans are profoundly polarized, the process of examining deeply held public values can enable the discovery of common ground. According to Pew Research, few Americans support the use of AI to advise people about faith in God, matchmaking, or governing the country. These areas of consensus, however narrow, provide a foundation for policy.

The challenge is that AI ethics principles can only reflect more than just the discourse of experts and computer scientists by being debated, hammered out, implemented, and modified in actual communities, governments, and businesses. Decisionmakers can get a head start by building upon public values, which have already emerged from iterative processes of trial, implementation, and deliberation embedded in law, regulation, and cultural practice.

The path forward requires moving beyond abstract principles and engaging the public values that communities have already developed over generations. Only then can AI governance become something that feels legitimate and reflects what people actually want from the technology.