From Adaptive to Automated Ecosystem Management
Artificial Intelligence (AI) is dramatically changing our capacities to collect, make sense of, and act upon, large, diverse and evolving datasets. The power of AI systems lies in their architecture, with learning frameworks, models, data and automation together establishing closed loops between insights and actions. This conceptually resembles the idea behind Adaptive ecosystem management (AM), which builds on repeated data collection, modelling, forecasting and decision to continuously improve knowledge and goal fulfilment. In this paper, we ask how AI can help address existing gaps in AM. We show that AM increasingly uses AI technologies, but implementation has generally been piecemeal and focused on specific problems, rather than used as a comprehensive and integrated technology for systems change. We suggest that AI can support a next generation of iterative, self-learning, real-time automated AM systems. To reach that goal, we present a development roadmap and discuss implementation risks and uncertainties.
Jonas Hentati-Sundberg, Henrik Österblom, Olof Mogren, Michael Dietze