Governance-Embedded AI Drought Early Warning Systems for Kenya's ASALs Using CAMEAL and Human Decision Layers

This study develops and evaluates a governance-embedded AI drought early warning system for Kenya’s ASALs using climate, vegetation, and food security data integrated through an R-based analytical pipeline. The research compares baseline machine learning models with governance-enhanced systems incorporating CAMEAL, HITL/HOTL oversight mechanisms, and MEAL feedback structures to assess how governance influences drought prediction, decision quality, accountability, and system stability. Large Language Models (LLMs) are used only as development-support tools for preprocessing and R script generation, and are excluded from operational decision-making.