As artificial intelligence (AI) becomes increasingly integrated into spatial planning and governance, a central challenge is translating AI-enabled technical capabilities into institutionally usable spatial decision-making capacity. Taking territorial spatial governance (TSG) in China as a context-specific governance setting, this study combines bibliometric analysis and structured review to examine its knowledge evolution, technical pathways, and spatial decision-making constraints. The results show that: (1) research has progressively shifted from digital support and spatial monitoring toward intelligent analysis and governance-oriented applications; (2) four recurrent pathways of AI embedding are synthesized from the literature—sensing and identification, analytical simulation, decision support, and platform coordination; (3) the translation of these technical capabilities into spatial decision-making capacity is constrained by heterogeneous data foundations, limited model interpretability and transferability, institutional misalignment, and uneven organizational implementation; and (4) spatial decision-making capacity emerges through the interaction of technical pathways and constraint mechanisms under the multi-scale, differentiated, and rule-bound spatial logic of TSG. These findings provide an analytical basis for understanding how AI-enabled technical capabilities are translated from technical feasibility into governance usability under institutionally and spatially specific conditions.

Agricultural land systems globally face escalating pressures from rising food demands, climate change, environmental degradation, and biodiversity loss. China, as a critical case, exemplifies the urgent need for strategies that reconcile food security with ecological sustainability. Here, we demonstrate that adopting a systematic approach to spatially allocate existing land policy tools—such as cropland reforestation, agricultural intensification, non-grain cropland restoration, and agricultural expansion—has the potential to simultaneously achieve multiple sustainability goals. Using a predictive model based on a socio-ecological-technical framework and machine learning, we evaluated the outcomes of six counterfactual scenarios for China’s agricultural land-use transitions at the county level. Results indicate that under a maximum land-sparing scenario (maximizing intensification of exist cropland, restoring unstable cropland, and maintaining non-grain cropland), compared to the 2020 baseline, China could increase grain output by 8%, reduce crop carbon emission intensity by 1%, enhance carbon sequestration by 63%, while substantially mitigating biodiversity loss across key taxa. However, the spatial distribution of land policy tools remains uneven, leading to varying types and degrees of trade-offs across specific counties under any given scenario. This highlights the critical need for coordinated national leadership to achieve sustainable objectives at a broader scale, offering valuable insights for global land-use transitions.

Decoupling the drivers of cropland productivity is critical for food security, yet the non-transferability of mechanisms across spatial scales hinders consensus. Productivity is an emergent property of a social-ecological system (SES). Here, we deconstruct the drivers of cropland productivity change across China (2001–2020) using a multi-scale analytical framework that integrates machine learning and causal inference. Our analysis reveals three key findings. First, productivity is co-determined by a nexus of biophysical properties (24.2% relative importance), urbanization (17.4%), and cropland-use intensity (16.2%), with pronounced non-linear interactions. Second, the dominance of these drivers is scale-dependent, shifting systematically from macro-scale biophysical processes to micro-scale socioeconomic behaviours. Third, external macro-scale disturbances primarily exert their influence through indirect micro-scale pathways; for instance, urbanization’s main impact is not direct land conversion, but the suppression of cropland use intensification, making it the dominant negative driver in 89% of counties. These findings demonstrate that cropland productivity dynamics are governed by a network of multi-scale, indirect causal mechanisms. Effective governance requires a systems-based approach that moves beyond single-factor interventions to manage these complex interactions.