Research Summary: Beyond the Townhall: Spatial Anchoring and LLM Agents for Scalable Participatory Urban Planning

Recent research explores the application of Large Language Models (LLMs) in public participation, specifically in urban planning contexts. An experiment demonstrated that immersive information delivery, such as narrated 360-degree video with spatial audio and anchoring, significantly improved participants' recall and influenced the nature of their interactions with LLM agents. This approach led to more detailed and positive feedback, suggesting a pathway for more effective public engagement processes.

Why it matters

This research highlights the potential for technology, specifically immersive experiences combined with AI, to transform public engagement and participatory processes. Improving how information is conveyed can lead to more informed public input, fostering greater buy-in and more robust decision-making in large-scale initiatives like urban planning and sustainability projects.

Key insights

• LLMs are being investigated for their utility in public participation scenarios, particularly in urban planning.
• The way citizens encounter information prior to engaging with LLMs impacts the quality and nature of their subsequent interactions.
• An experimental study (N=200) compared information delivery methods for a sustainability urban planning project: text/images versus narrated 360-degree video with spatial audio and spatial anchoring.
• The immersive information condition (360-degree video, spatial audio/anchoring) enhanced participant recall of project details.
• Participants exposed to the immersive condition provided different feedback to LLM agents, including more frequent naming of beneficiaries, more project-related questions, and increased positive feedback.
• The study conceptualizes LLMs as dialogue agents within public participation frameworks.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2604.16348