Drivers of advisory chatbot adoption in smallholder agriculture: A MICMAC analysis of farmer intermediary perceptions in the Sundarbans, India

In low-literacy, climate-vulnerable agricultural settings, smallholder farmers rarely evaluate new technologies independently; adoption is mediated through trusted institutional intermediaries. However, the systemic factors driving intermediary-perceived adoption of AI advisory tools remain unmapped. This study applies the Matrice d'Impacts Croisés-Multiplication Appliquée à un Classement (MICMAC) method to examine the systemic drivers of the adoption of generative AI-based agricultural chatbot s , namely, Krishi Dishari, based on the responses of thirty Gram Unnayan Karmi (GUK) intermediaries in the Sundarbans, India, a region facing salinity intrusion and extreme climate challenges. Using a structured bilingual influence matrix survey, the study mapped and classified twelve socio-technical variables by their driving and dependence power. The results indicate that institutional trust is a critical systemic linkage variable, with smartphone access serving as the primary technical driver and social learning emerging as the dominant diffusion mechanism. Furthermore, native language comfort (Bengali) emerged as a key transmission driver, confirming that linguistic accessibility is a prerequisite rather than an enhancement for technology adoption in low digital-literacy contexts. Consequently, this study offers a framework, although transferability warrants exhaustive validation, indicating that AI-based advisory services in vulnerable agricultural settings depend on institutional trust and social learning, not only on technological innovation.