Irrigation with Artificial Intelligence: Problems, Premises, Promises

Protagonists allege that artificial intelligence (AI) is revolutionising contemporaneous mindscapes. Here, we authoritatively review the status quo of AI and machine learning application in irrigated agriculture, evaluating the potential of, and challenges associated with, a wide range of existential AI approaches. We contend that aspiring developers of AI irrigation systems may benefit from human-centred AI, a nascent algorithm that captures diverse end-user views, behaviours and actions, potentially facilitating refinement of proposed systems through iterative stakeholder feedback. AI-guided human–machine collaboration can streamline integration of user needs, allowing customisation towards situational farm management adaptation. Presentation of big data in intuitive, legible and actionable forms for specialists and laypeople also urgently requires attention: here, AI-explainable interpretability may help harness human expertise, enabling end-users to contribute their experience within an AI pipeline for bespoke outputs. Transfer learning holds promise in contextualising place-based AI to agroecological regions, production systems or enterprise mixes, even with limited data inputs. We find that the rate of AI scientific and software development in recent times has outpaced the evolution of adequate legal and institutional regulations, and often social, moral and ethical license to operate, revealing consumer issues associated with data ownership, legitimacy and trust. We opine that AI has great potential to elicit sustainable outcomes in food security, social innovation and environmental stewardship, albeit such potential is more likely to be realised through concurrent development of appropriate ethical, moral and legal dimensions.