A Prospectus on Generative Artificial Intelligence in Marine Ecosystem Modelling

ABSTRACT Marine ecosystem modelling faces increasing demands for rapid development and deployment to address urgent environmental challenges, yet technical complexity and time‐intensive processes often constrain timely insights for management decisions. This prospectus synthesises current applications and outlines future research directions for integrating Generative Artificial Intelligence (GenAI) into marine ecosystem modelling while maintaining scientific rigour. I present a structured framework for integrating GenAI across eight interconnected components of the modelling cycle: model scoping, data gathering, conceptual framework development, model development, model execution, validation and calibration, reporting and stakeholder engagement. Through analysis of current applications and emerging research, I demonstrate how GenAI can automate routine tasks, democratise access to sophisticated modelling approaches, and improve model quality. Achieving success will require overcoming persistent challenges, including data limitations, institutional barriers and ethical concerns. I propose a research agenda addressing three streams: capability assessment to systematically evaluate GenAI's potential in marine ecosystem modelling; avenues for ensuring scientific integrity and reliability; and socio‐technical integration to address ethical and institutional challenges. While GenAI offers the potential to enhance modelling, a human‐centered approach is essential, where GenAI augments, rather than replaces, human expertise in model validation, interpretation of results and ensuring sustainable management outcomes. To support readers new to this space, a primer in the supporting information outlines practical considerations for accessing GenAI tools, from cloud‐based services to locally‐run models and their implications for privacy, reproducibility and computational requirements.