Artificial intelligence in sustainable agriculture: Towards a socio-technical roadmap
Peiqian Wu, Yuxin Zhong
Abstract
Agriculture stands at an important juncture, tasked with meeting global food demand amidst unprecedented climate and resource pressures. Artificial Intelligence (AI) has emerged as a transformative force. Research from 2021 to 2024 demonstrates a maturation characterized by the transition from experimental proofs-of-concept to field-validated applications exhibiting robust performance, including high accuracy (often exceeding 90% in detection tasks) and measurable resource efficiency gains in precision management, robotics, and predictive analytics. However, the translation of this technical promise into equitable, scalable, and sustainable agricultural systems is increasingly bottlenecked by critical socio-technical and ethical challenges. This review synthesizes key technological advancements from the past four years and argues that the next frontier of innovation requires a pivot from a purely techno-optimist focus on algorithmic performance to a socio-technical one. We analyze how emerging paradigms, namely Agricultural Digital Twins and Human-AI Collaboration, represent this shift toward integrated, human-centric systems. Critically, the success of these paradigms and the responsible deployment of AI in agriculture hinge on proactively addressing systemic issues of algorithmic bias, data sovereignty, and the socio-economic impacts on the agricultural workforce. By synthesizing these technical, ethical, and policy dimensions, we propose a strategic roadmap for research, development, and governance to guide the next phase of innovation, ensuring that AI fosters a more productive, resilient, and just food future.