Envisioning transformative solutions for climate risks demands innovative communication tools that can integrate perspectives from civic institutions, scientists, and practitioners. However, facilitating interdisciplinary conversations around the complexity of these issues and their solutions requires sensitivity to varied technical languages, terminologies, and conceptual frameworks. The tangibility of board games coupled with the visualization power of simulations offers a promising, yet underutilised arena for integrating the varied perspectives when addressing environmental challenges. This project outlines a novel approach leveraging computer vision and AI/ML with a socio-environmental board game to significantly enhance the communication among the interdisciplinary stakeholders (i.e., players), and contribute to the development of human-in-the-loop AI systems for environmental challenges.
We propose to use advanced AI/ML algorithms operating on video data, to detect, analyze and visualize the real-time state of the environmental board game, ”FutureScape.” FutureScape is a serious game designed to elicit human behavior in the face of climate change, simulating real-world policy deliberations under conditions of imperfect information, limited resources, and pressing time constraints. Our proposed approach (1) draws on the potential of computer vision techniques for detecting players actions and consequences for recordkeeping and analysis, (2) uses ML to extract intelligence in real-time that would be presented to the stakeholders to assess immediate and long-term costs and benefits, and (3) integrates the intelligence with established climate models to simulate the long-term spatio-temporal, cascading consequences of the collaborative actions. By visualizing the direct link between their decisions and environmental impacts, players will develop a deeper, more salient understanding of complex environmental systems, identify variables that need to be prioritized in every scenario, and explore multiple solutions within their constraints. Such experiential environments used in education and planning activities have been seen to facilitate decision-making for both experts and novices by simplifying complex environmental issues into accessible simulations, providing tangible and visual representations of impacts, and enabling active experimentation and learning through direct experience within a safe, interactive micro-world. The long-term vision is to utilize this unique dataset of human-decision making in socio-environmental contexts to train a custom AI system that works hand-in-hand with the players. This project constitutes the first step towards a contribution to human-AI teaming in environmental research, necessary to build a climate-ready nation and enhance national security.