Strategically constructed narratives on artificial intelligence: What stories are told in governmental artificial intelligence policies?

Abstract

What stories are told in national artificial intelligence (AI) policies? Combining the novel technique of structural topic modeling (STM) and qualitative narrative analysis, this paper examines the policy narratives in 33 countries’ AI policies. We uncover six common narratives that are dominating the political agenda concerning AI. Our findings show that the policy narratives' saliences vary across time and countries. We make several contributions. First, our narratives describe well-grounded, supportable conceptions of AI among governments, and show that AI is still a fairly novel, multilayered, and controversial phenomenon. Building on the premise that human sensemaking is best represented and supported by narration, we address the applied rhetoric of governments to either minimize the risks or exalt the opportunities of AI. Second, we uncover the four prominent roles governments seek to take concerning AI implementation: enabler, leader, regulator, and/or user. Third, we make a methodological contribution toward data-driven, computationally-intensive theory development. Our methodological approach and the identified narratives present key starting points for further research.