Abstract for: Exploring AI Performance In Dynamic Decision-Making: Case Study on Fishery Management

Overfishing and mismanagement pose significant threats to fisheries sustainability, often driven by cognitive biases and lack of system and feedback understanding. This study aims to compare human-based fishery experiments and AI-integrated model output and understand whether AI addresses human cognitive limitations found in the original fish experiment. System Dynamics approach is employed in this study by replicating Moxnes’ fishery experiment model in Insight Maker to analyze financial and resource biomass performance under different decision-making approaches. AI is then incorporated to make sequential decisions on fleet expansion and lay-up. In the original fishery experiment, human participants overexpanded fleets, leading to sharp biomass decline and financial instability. After incorporating AI, the results demonstrated that AI exhibited more cautious and adaptive decision-making than human participants, avoiding over-expansion, maintaining financial stability, and demonstrating an ability to recognize feedback patterns that lead to long-term sustainability. Future research should explore AI’s performance under different parameter settings, its sensitivity to external inputs, and its ability to define objectives autonomously. Additionally, it also explores if AI can used to optimize decision-making in resource management and to test policy effectiveness, simulating how human-like decision-makers may respond to new regulations.