How water is used in farming plays a key role in understanding the difficulties of managing groundwater resources. Therefore, it is important to recognize and model how farmers behave as decision-makers—since their choices affect the environment, as well as officials and other stakeholders. The Ardabil Plain aquifer, which is the main source of water for farming in the area, has seen a major drop in its groundwater levels in recent years due to over-extraction. This study presents a combined framework that uses agent-based modeling (ABM) and artificial intelligence (AI) to simulate how farmers act when using groundwater and to assess how their actions impact the aquifer. Machine learning and neural networks are used to model how farmers make decisions. When modeling their choice of which crops to grow, the dryness conditions in the coming water year are taken into account. For this, a climate-based drought index is forecasted using rainfall data from weather prediction models—after these data have been downscaled and corrected for errors. By building this ABM-AI combined model, we analyze how farmers interact with each other and how willing they are to protect groundwater. We also evaluate how different management plans would affect farmers' behavior and the water conditions in the region. The findings show that good management plans can cut groundwater pumping from wells by 8.8% on average. In addition, these plans lead to more land being used to grow crops that need less water and are of strategic importance to the Ardabil Plain.