How Citizen Assemblies Can Help Create Inclusive Tech Policy
Sam Wollenberg - September 15, 2023
In recent years, the rapid advancement of AI-enabled technology has outpaced the ability of global lawmakers to regulate its wide-ranging impacts on individuals, communities, and democratic institutions. This widening gap prompts crucial questions about not only the development of effective policies but also the role of citizens in shaping these frameworks. Answering these questions will help mitigate the negative experiences faced by individuals and communities using these technologies, including but not limited to women, trans, and POC.
In this context, a renewed focus on citizen assemblies has emerged and presents an exciting opportunity to create more participatory, equitable, and responsible technology policy. These assemblies aim to democratize the policy-making process by inviting a diverse and representative group of citizens to partake in expert-informed deliberations on regulatory strategies. Not only do these forums enrich the content of policies, but they also bolster public trust in the legislative process, especially as governments scramble to understand and govern swiftly evolving technologies.
The process of participant selection in citizen assemblies is typically guided by the principle of 'sortition.' This involves using existing data sources, like census records, to randomly invite a cross-section of citizens. Multiple channels, including email, phone, and mail, are employed to maximize inclusivity. The objective is to obtain a sample that is sufficiently large and diverse to represent the broader population.
The number of invitations often depends on the population size of the country or community you are seeking to engage with. Facilitators like Democracy Next recommend a wide range between 2,000 and 30,000 in this first round of sortition depending on policy objectives. Once invitees confirm their willingness to participate, a second layer of sortition is performed, this time controlling for age, sexual and gender identity, location, and socio-economic status to create a final group of proportionally representative participants.
The efficacy of citizen assemblies hinges on several vital elements: subject-matter experts, citizens from diverse professional backgrounds, and skilled facilitators. These components become particularly indispensable when navigating intricate issues like generative AI and Large Language Models (LLMs). The complexity and ethical ramifications of such technologies necessitate a multi-dimensional approach to policy-making, requiring expertise for technical depth, citizen input for social context, and adept facilitation to synthesize these diverse perspectives into actionable policy. By designing representative citizen assemblies, groups seeking a more responsible technology landscape can leverage this process to put power directly in the hands of those most impacted.
One recent example is the Citizens Biometric Council which was established in 2020 by the Ada Lovelace Institute in the UK to inform legislation for the collection and use of biometric data. This council produced a range of suggested standards that biometric technology regulation should cover to meet participant expectations and found that engaging with and embedding underrepresented or marginalized voices in the council to be essential for the development of an informed policy.
Similarly, the Citizens' Assembly on Democratic Expression in Canada articulated public appetite for urgent, far-reaching regulations on issues like online hate speech and exploitative business models. These assemblies represent an evolution in public consultation methods—offering more focused, informed, and actionable discussions compared to traditional forums or town halls.
Fig. 1 provides an overview of the citizen assembly process from Democracy Next, an international non-profit, non-partisan research and action institute.
In contrast, the current landscape often relegates citizens to engaging with tech policy through social media, platforms ill-suited for nuanced deliberation. These platforms not only prioritize ad revenues but also yield disparate user experiences. For instance, a recent Canadian survey revealed that women, trans individuals, and people of color face higher rates of online harassment, highlighting the need for policy to be as varied and nuanced as the experiences it seeks to regulate.
Meanwhile, new issues are emerging surrounding the potentially disruptive effects of AI. In the last year, the illegal and non-consensual use of artists’ creative works to train AI is already having potentially significant ramifications on livelihoods. As AI becomes increasingly integrated across sectors, including the voices of those most impacted by emerging technologies is essential to understanding the social, economic, and professional impacts that policy must address. Citizen assemblies provide effective forums to weigh perspectives and evidence and provide an alternative to the disjointed and attention-grabbing model of our current information ecosystem.
Such mechanisms are even more important now given the increasing rates of political polarization among Canadians which is impacting conversations and debates about pressing issues in our current digital environments. Citizen assemblies have the potential to distribute decision-making power in the regulatory processes via a controlled and collaborative environment where both public and expert opinions are considered. Canadian policymakers and politicians need to take serious note of this inability to respectfully deliberate and should strongly consider leveraging citizen assemblies as a way to improve policy and restore trust in democracy’s ability to grapple with complex challenges.
