Artificial intelligence for climate-smart agriculture: Bridging science, policy, and practice - A narrative review

Abstract

Climate change poses a serious challenge to global food security, where agriculture will need to be highly productive, adaptive to climate change, and also less emission of greenhouse gases into the atmosphere. Climate-smart agriculture (CSA) plays a central role in addressing these challenges, but its popularity, especially with smallholder farmers in low- and middle-income nations is hampered by the lack of data, decision support, and adaptability. The use of artificial intelligence (AI) is becoming a revolutionary technology that can be used to bridge scientific knowledge, policy formulation, and real-world farming practices to boost CSA realization. This narrative review was based on the peer-reviewed and grey literature published in last two decades with the aim of assessing the AI application in the main fields of CSA, such as crop yield prediction, precision irrigation, pest and disease surveillance, soil carbon monitoring, as well as digital advisory services. It synthesized data from 177 sources from various research databases (Web of Science, Scopus, Google Scholar) identified technological trends, policy frameworks, and an adoption barrier model. Additionally, it outlined a future research roadmap and provides specific policy and practice recommendations. The review concluded that despite significant potential in all the studied CSA domains, with 15-30% improvements in yield prediction accuracy and 20-40% reductions in water consumption in controlled environments (single-study estimates reported without confidence intervals; see Limitations), the adoption of AI applications by smallholder farmers are crippled by the fragility of data ecosystems, infrastructure gaps, limited institutional capacity, and systematic inequities in the global AI innovation ecosystem. The review found that to achieve the transformative potential of AI in CSA, there is an urgent need to re-orient the research investment, governance structures and institutional structures to the complex, data-poor realities of tropical smallholder farming systems.