Artificial Intelligence Governance as Policy Discourse: A Computational Analysis of Official Indian Policy Documents

Abstract

AI governance in India is usually examined through legal, institutional and strategic perspectives; however, limited attention has been paid to how governance priorities are communicated and how policy language shapes authority, responsibility, and inclusion. This study thus approaches governance as a policy discourse rather than implementation outcomes through an interpretive-computational research design that combines computational text mining with discourse-oriented interpretation. A dataset of national-level AI policy documents was curated through systematic retrieval and screening. Policy sections were analysed using semantic modeling, framing analysis, and lexical polarity analysis with emphasis on interpretive analysis and institutional context rather than predictive modeling or frequency-based measures. The findings show that AI governance discourse in India is institutionally coordinated and development oriented. Innovation, economic growth and institutional coordination emerged as dominant themes while ethics, risk, and data-protection concerns appear primarily within regulatory contexts. In framing analysis, opportunity-oriented domains dominate regulatory domains. Risk and caution are addressed primarily within regulatory domains. This study demonstrates how AI governance operates through discourse and rhetorical framing, and not only through regulation. It positions computational text mining as a research design, offering a scalable approach for analysing the social policy dimensions of emerging technology governance.