Can Machine Learning Support Planning for Equitable Green Infrastructure? A Perspective on Opportunities, Risks, and Ethical Pathways

Urban green infrastructure (GI) is widely promoted for cooling, stormwater regulation, biodiversity support, and human health benefits; however, these benefits remain unevenly distributed across communities with different socio-economic status. Machine learning (ML) is increasingly used in GI planning and design through high-resolution mapping and demand and exposure modelling, enabling planners and landscape architects to explore scenarios and assess alternative solutions based on the benefits generated. However, ML risks generating or perpetuating spatial injustice through biassed training, opaque optimisation priorities, and epistemic exclusion of Indigenous and local knowledge when models are not developed and applied transparently and collaboratively. This perspective discusses recent GI and ML trends and debates to: (i) clarify how ML can support equity-oriented GI planning; (ii) identify technical and socio-economic risks; and (iii) outline ethical and governance pathways supportive of legitimate and accountable GI planning. We argue that ML should be treated as a component of socio-technical governance rather than a neutral technical tool and therefore should be applied through collaborative design and periodic re-evaluations.