Bridging Deliberative Democracy and Deployment of Societal-Scale Technology

Current AI safety processes lack democratic legitimacy: The paper argues that existing safety measures for large language models are insufficient because they don't involve meaningful input from the diverse users and stakeholders affected by these technologies, raising questions about who gets to decide how these systems are deployed in society.

Deliberative democracy should guide AI deployment: Rather than relying solely on technical experts and companies, the author advocates for structured deliberative processes where users and stakeholders can collectively discuss and decide important questions about LLM safety and appropriate deployment contexts.

LLMs qualify as "societal-scale technology": The paper establishes that large language models have reached a level of societal impact comparable to other major technologies (like electricity or transportation systems), meaning they warrant the same democratic oversight and public deliberation processes we apply to other critical infrastructure.

New design work is needed to translate public input into practice: The author identifies a gap that requires both policy design (determining how to conduct public deliberation) and technical design (creating interfaces and features that actually incorporate deliberative outcomes into development processes).

HCI professionals have a key role to play: The paper calls on human-computer interaction designers specifically to help bridge the gap between democratic deliberation and technical implementation, positioning them as essential to making AI governance more participatory and legitimate.