Can't stop the hype: scrutinizing AI's realities

Recently, we have been inundated with reports about the revolutions brought about by AI. However, much of our knowledge about these ‘revolutions’ arises from media sensationalism and advertising clamor – from AI hype. Therefore, to understand the societal impacts of AI, we must first grapple with its hype and ask: What truly transpires in the ‘revolutions’ allegedly brought about by AI? What role does hype play in them? This paper addresses this need by offering a methodological framework for studying AI, a way to scrutinize AI hype while acknowledging its power. I begin by presenting the STATE typology of AI hype that differentiates between five distinct types of hype. I then offer six heuristics to identify and explore AI hype. Lastly, I argue that holistic empirical research can leverage the STATE typology and its heuristics into a robust methodological design for scrutinizing AI-driven phenomena and their hype.