Artificial intelligence and digital phenotyping shape the future of plant breeding: evolution, challenges, and a roadmap from AI breeding 1.0 to 2.0

Abstract

Traditional breeding is approaching its efficiency limit in addressing global food security challenges, climate change, and multi-dimensional information integration. Artificial intelligence (AI) and digital phenotyping have become core driving forces in data-driven breeding. This review systematically elaborates the evolutionary roadmap from AI breeding 1.0 , which relies on traditional phenotypic data, to the emerging paradigm of AI breeding 2.0 . We clarify the core differences between these two paradigms, dissect the “data divide” in the transition process, and summarize the integrated technical framework underlying AI breeding 2.0 . The central argument is that the core limitation of AI breeding 1.0 lies in data inadequacy rather than constraints on algorithmic capacity. The transition to AI breeding 2.0 relies on standardized high-throughput digital phenotyping, multi-modal data integration, and closed-loop data-centric breeding systems. Finally, we propose four priority actions to promote the widespread adoption of data-driven breeding and provide an actionable roadmap for enhancing global crop improvement efforts.

Code and data availability

This is a perspective/review article on AI and digital phenotyping in plant breeding. It contains no original plant-phenotyping measurements, datasets, images, sensor data, or author analysis code. All tools and platforms mentioned (OpenPheno, TasselNetV4, FoMo4Wheat, TraitDiscover, TerraSentia, etc.) are cited prior/외

No evidence-backed public reproduction asset is currently recorded.