LLMs and Participatory Planning: A New Approach

Socio-environmental planning under deep uncertainty requires an initial phase of problem conceptualization, often based on participatory modeling. These models translate stakeholder descriptions into quantitative models, a complex and time-consuming process.

A recent study proposes a workflow that leverages large language models (LLMs) to automate this initial phase. The goal is to identify the essential model components from stakeholder problem descriptions, exploring different perspectives and assembling a unified model.

Workflow Details

The workflow involves using LLMs to:

  • Identify key model components from stakeholder descriptions.
  • Explore different perspectives on the problem.
  • Assemble the components into a unified model.
  • Implement the model in Python through successive iterations.

Researchers tested the workflow using ChatGPT 5.2 Instant on two socio-environmental planning problems, obtaining acceptable outputs after a few iterations with human verification and refinement. The results suggest that LLMs can be an effective tool for facilitating participatory modeling in the problem conceptualization phase.