This thesis explores the multi-dimensional interplay between natural resource wealth, governance, and democratic development using machine learning techniques. The methodology of this research employs the Random Forest algorithm on comprehensive country-level data – including predictors using resource rents, exports, GDP per capita, corruption perceptions, and education levels – in order to investigate how resource abundance, often associated with the “resource curse” and rentier state dynamics, influences governance outcomes in nations with varying degrees of transparency and institutional strength.

Traditional models suggest that resource wealth increases the likelihood of autocratic political outcomes; however, this research uses AI to confirm this relationship and takes it a step further to examine how governance quality, particularly transparency, plays a role. The machine learning analysis reveals that transparency significantly enhances democratic resilience in resource-rich countries. These findings indicate that strong governance and transparent institutions can mitigate the democratic risks posed by resource dependency, challenging deterministic interpretations of the resource curse. Ultimately, the study contributes to the literature by highlighting the crucial role of institutional quality and transparency in fostering democracy within resource-dependent states, offering valuable insights for policymakers seeking to break the cycle of rentierism and autocracy.