Democracy at risk as policymakers overlook citizen input on AI regulation

Governments often release consultation materials filled with technical jargon, making them inaccessible to people without advanced digital or policy knowledge. In some cases, public submissions required navigating complex online forms or writing formal documents, barriers that excluded the average citizen. The researchers argue that true participation requires not just open calls but proactive outreach, plain-language communication, and inclusion of those typically left outside the digital sphere.

A new study warns that governments worldwide are failing to meaningfully include citizens in discussions about artificial intelligence (AI) policy, despite claims of public engagement.

The research, titled "Lost in Translation: Policymakers Are Not Really Listening to Citizen Concerns About AI," reveal that public consultations in Australia, Colombia, and the United States, intended to shape national AI strategies, largely excluded ordinary citizens, leaving critical voices unheard in shaping the technology's future governance.

Governments seek public input, but few are heard

The study investigates how governments attempt to involve citizens in AI policymaking and how effective these processes truly are. Through a comparative analysis of three national consultations, the authors show a troubling gap between political rhetoric and real participatory practice.

In Australia, the Department of Industry, Science, and Resources launched one of the most extensive national consultations, receiving around 510 submissions. Yet most participants were from corporate or institutional backgrounds, and the results showed limited reflection of community-level concerns. While the process included town halls and roundtable discussions, these events did not reach underrepresented populations or the general public.

In the United States, the National Telecommunications and Information Administration (NTIA) received 326 responses in its public comment process. However, the analysis found that most inputs came from policy insiders, think tanks, and private sector representatives. Ordinary citizens were scarcely involved, and there was little evidence that public submissions influenced the final framework.

Colombia showed the weakest engagement, with only 73 responses gathered by the Ministry of Science, Technology, and Innovation. The consultation was short, open for just one week, and poorly publicized, resulting in an unrepresentative sample dominated by academic participants.

Across all three cases, participation rates were strikingly low, less than one percent of the population. Moreover, none of the consultations produced transparent explanations of how citizen feedback was used in policymaking. The authors describe this as a critical flaw in democratic AI governance: governments appear to invite dialogue, yet decisions are still shaped by narrow, expert-driven perspectives.

The engagement gap: Tokenism over true participation

The authors used the International Association for Public Participation (IAP2) Spectrum to assess the depth of citizen involvement in each process. The framework distinguishes between minimal consultation and genuine empowerment. None of the countries studied achieved the higher tiers of engagement, such as collaboration or citizen-led decision-making. Instead, most efforts fell into the lower categories: informing and consulting.

This limited approach reflects what the authors call policy tokenism, a superficial effort to create the appearance of inclusion without granting citizens any real influence. The paper emphasizes that this approach fails to capture the diversity of public sentiment about AI, particularly from marginalized communities most affected by automation, surveillance, and algorithmic bias.

Another challenge highlighted in the study is accessibility. Governments often release consultation materials filled with technical jargon, making them inaccessible to people without advanced digital or policy knowledge. In some cases, public submissions required navigating complex online forms or writing formal documents, barriers that excluded the average citizen. The researchers argue that true participation requires not just open calls but proactive outreach, plain-language communication, and inclusion of those typically left outside the digital sphere.

The study finds that these shortcomings undermine both policy quality and public trust. When citizens see that their opinions do not matter, they are less likely to engage in future consultations. This dynamic, the authors suggest, risks entrenching a cycle of disengagement in which governments increasingly rely on experts and corporations while losing touch with public priorities.

Rebuilding trust through inclusive AI policymaking

The researchers argue that reversing this trend requires a fundamental rethinking of how governments approach AI governance. They propose eight concrete steps to make public engagement more meaningful and representative.

First, the study calls for strengthening AI literacy so that citizens can participate knowledgeably in debates about data, automation, and ethics. Without foundational understanding, public discussions remain dominated by elite voices. Second, governments should monitor and publicize how public feedback is used to shape policy outcomes, creating accountability mechanisms.

The researchers also call for diverse communication strategies, including outreach through television, radio, and social media to reach rural and underserved populations. They suggest hosting online AI town halls and community-based forums that allow two-way discussions rather than one-way surveys.

In addition, innovative engagement formats, such as AI policy hackathons or citizen juries, could encourage practical, solution-oriented participation. Governments should also ensure the inclusion of marginalized or offline communities, not only digital participants.

The authors argue that transparency must become a core pillar of AI policymaking. Citizens should be able to trace how their feedback affects legislation, white papers, and regulatory actions. Finally, they propose leveraging AI itself as a democratic tool: large language models and swarm intelligence systems could analyze vast public inputs, synthesize themes, and help policymakers identify genuine societal concerns.

These reforms, the study concludes, are vital to maintaining democratic legitimacy as AI systems increasingly shape economic, political, and social life. Without inclusive participation, governments risk building AI policies that reflect the interests of the few rather than the values of the many.