Improving plant breeding through AI-supported data integration

Abstract

Integrating, learning from, and predicting using vast datasets from various scales, platforms, and species is crucial for advancing crop improvement through breeding. Artificial intelligence (AI) is a broad category of methods, many of which have been used in breeding for decades. Recent years have seen an explosion of new AI tools (or old ones at new scales), with exciting applications, both demonstrated and potential, to improve or maybe even revolutionize plant breeding! Example use cases and data types included data mining, phenotyping, monitoring, genetics, multi-omics, environment, management practices, cross-species inference, sustainability, economics, and many others. Improvements in these areas could increase predictive accuracy for plant traits, thereby expediting breeding cycles and optimizing resource management. Aside from improving predictions, AI methods can potentially enhance biological inferences and enable more informed approaches to areas like gene discovery, gene editing, and transformation. At the same time, AI is not going to solve every breeding challenge, and studies so far have shown mixed results depending on the application, dataset, and other factors. AI continues to transform plant breeding, yet its full potential remains unclear, with many possibilities still to be realized. This review explores the transformative potential of AI in plant breeding with a particular focus on its ability to integrate the many diverse streams of data involved. Success in this would open opportunities to improve crop resilience, yield, and sustainability, thus supporting global food security and inspiring the next generation of plant breeding technologies.