<p>In 2020, the State of California passed legislation requiring law enforcement agencies to share their policy manuals online to encourage “meaningful public input” on police policy. The documents, though, are written to reduce the legal liability of law enforcement rather than enhance public understanding of law enforcement policies. It is essential, then, to address what it means to provide access if the goal is to inform the public. Making the documents merely available and queryable is not enough. Instead, we offer a plural context discovery system that surfaces core concepts within the policy manuals and draws them into public discourse – fundamentally reimagining how citizens engage with and shape policy understanding. This project proposes a paradigm shift from passive information access to participatory knowledge construction. We historicize concepts, acknowledging that they change over time, that they are formed in social contexts, that interpretation of their meaning is often contested, and that evidence to support interpretation can be contradictory. We demonstrate that words presented as entities within a knowledge graph, connected in a network of temporal relationships, take on the ontological complexity of concepts. Too often, knowledge graphs present singular “weak” definitions that do not allow for rich and nuanced engagement with concepts. We propose placing generative AI in an adjunct role within a plural context discovery system that is designed to encourage critical engagement and reflection. Plural context discovery removes the AI agent from the role of information arbiter and focuses instead on its capacity to predict. In this case it is tasked to predict plural viewpoints when a concept represented in a knowledge graph references a limited or static point of view or is poorly supported by evidence. The design goal is to create a participatory system in which the public is empowered to generate, share, and transform knowledge about police policy.</p>