摘要
Abstract Forests occupy a pivotal position in addressing the intertwined crises of climate change, biodiversity loss, and sustainable resource use. Yet, the United Kingdom, one of Europe’s least-wooded and most nature-depleted nations, faces persistent barriers to resilient forest management: low forest cover, narrow species diversity, dependence on timber imports, and fragmented ownership. Building on a national horizon scan conducted in 2022, which identified ‘catastrophic forest ecosystem collapse’ as the most pressing emerging issue, this paper explores how artificial intelligence (AI) can strengthen foresight and policy coherence. Although large language models and other AI tools were technically available during the 2022 horizon-scanning exercise, their use in environmental foresight was not yet methodologically normalized or institutionally validated. The absence of AI from that process should therefore be understood less as an omission than as evidence of a rapidly changing methodological context. We propose an AI-assisted horizon-scanning framework that combines AI-enabled signal detection and synthesis with expert interpretation, deliberation and judgement through six iterative stages: framing, signal discovery, expert engagement, synthesis, simulation, and reflexive learning. The framework operationalizes principles of Responsible Research and Innovation and Responsible AI to ensure transparency, accountability, and ethical governance while addressing the ‘data deserts’, institutional fragmentation, and limited place-sensitive decision-making that currently limit the deployment of AI in the forestry sector. A concept for a UK Forestry Foresight Lab is advanced as a practical institutional mechanism linking government, research, and landowners through federated data commons. The paper outlines how ethically-governed AI can transform horizon scanning from a static expert exercise into a dynamic, adaptive and place-sensitive foresight system, supporting sustainable forest governance across the UK’s devolved administrations.