����ֱ��

��

Communities across the United States face diverse water resource challenges requiring both rapid decision-making for urgent events and strategic long-term planning for chronic issues. Engineers and scientists rely on sophisticated models and simulations to understand these complexities; however, their outputs often remain inaccessible or impractical for real-world planning and decision-making processes. Artificial intelligence (AI) is increasingly proving valuable in enhancing decision-making across numerous fields. While machine learning algorithms and large language models dominate public discourse, other AI research domains offer significant potential to bridge the gap between advanced simulations and practical applications. Human-AI partnerships are particularly effective for addressing complex problems that exceed human reasoning capacity alone.

The AI-enabled Modeling Flagship project (AIM) exemplifies a human-centered approach, designing tools that fit decision-making workflows. The AIM project uses semantic AI to create recommender systems that suggest viable solutions backed by trustworthy data and bridges cross-disciplinary models and data with reusable tools and provenance tracking. Applications span hydrological models from coastal surge to flood engineering, and from drought response to subsidence impacts. To support the broad range of hydrological science and planning needs, our research team is linking cloud spaces with traditional supercomputing at scale, even as we customize problem framing interfaces that support instantiation of bespoke scenarios for multiple user audiences. Every advance leverages AI under the covers of our applications which is pivotal for developing intelligent decision support applications that enable researchers and communities to effectively respond to extreme events, protect vulnerable populations, and manage valuable natural resources. The integration of advanced AI with scientific data and action-oriented contexts will be a key area to watch in the coming years. Most importantly, finding AI-approaches that accelerate the spaces “in between” common scientific or analytical tasks can accelerate research to results.