Promising the future, encoding the past: AI hype and public media imagery

Abstract In recent years, “AI hype” has taken over public media, oscillating between sensationalism and concerns about the societal implications of AI growth. The latest historical wave of AI hype indexes a period of increased research, investment, and speculation on machine learning, centred around generative AI, a novel class of machine learning that can generate original media from textual prompts. In this paper, I dive into the production of AI hype in online media, with the aim of prioritising the normative and political dimension of AI hype. Formulating AI as a promise reframes it as a normative project, centrally involving the formation of public and institutional confidence in the technology. The production and dissemination of images, in this context, plays a pivotal role in reinforcing these normative commitments to the public. My argument is divided into four sections. First, I examine the political relevance of stock images as the dominant imagery used to convey AI concepts to the public. These stock images encode specific readings of AI and circulate through public media, significantly influencing perceptions. Second, I look at the dominant images of AI as matters of political concern. Third, as generative AI increasingly contributes to the production of stock imagery, I compare the epistemic work performed by AI-generated outputs and stock images, as both encode style, content, and taxonomic structures of the world. I employ an entity relationship diagram (ERD) to investigate the political economy of AI imagery in digital media, providing a snapshot of how AI hype is materialised and amplified online. With this study, I reaffirm AI’s normative character at the forefront of its political and ethical discourse.