A multi-source artificial intelligence framework for understanding climate effects on quality of agricultural products and supporting sustainable food systems

Quality of agricultural products (QAP) underpins global food systems, economic value, and public health. Climate change is altering QAP with significant implications for global food security. Understanding how climate shapes QAP is central to meeting rising demand for nutritious, flavourful, sustainable diets, and supporting food security. Artificial intelligence (AI) has accelerated progress in these areas through data fusion, non-linear modelling, and improved interpretability. However, increasingly complex climatic changes and intricate networks underlying QAP make it challenging to determine precisely how climate shapes QAP. Accordingly, this review synthesises three mainstream approaches to linking climate and QAP, i.e., data-driven methods, process-based crop models (PBMs), and manipulation experiments, with each approach having clear strengths and characteristic limitations. Furthermore, we show how AI already strengthens each method by fusing heterogeneous datasets, modelling non-linear relationships with interpretability tools, improving PBM inputs and structure, filtering output errors, and extracting informative signals from multi-omics to support causal tracing. Building on this assessment, we propose an AI-centred, multi-source integration framework that operationalises the fusion of these approaches to strengthen prediction, interpretation, and decision support. We contend that the integration of AI-enabled, multi-source approaches into agricultural practice could support climate-resilient agricultural development and provide new solutions to ensure global food security.