Inventing Rather Than Predicting: Generative AI, Anticipatory Anxiety, and a Research Agenda for the Futures

Generative AI has intensified a familiar orientation in modern institutions: the impulse to treat the future as something that can be forecast, priced, and governed through increasingly sophisticated prediction. Yet the dominant social response to current state-of-the-art AI deployments is not confidence in foresight but an unsettled anxiety that spans occupations, sectors, and public life. We argue that the anxiety is better understood as a consequence of anticipatory saturation - a condition in which predictive infrastructures multiply possible futures while simultaneously constraining agency in the present. Drawing on work on anticipation, sociotechnical imaginaries, and algorithmic governance, we suggest that prediction can become a technology of closure when it is treated as a substitute for deliberation. We then outline an alternative stance for Futures scholarship: shifting analytic and methodological emphasis from forecasting what AI will do to inventing, testing, and contesting what AI-enabled futures could be. We propose a concise research agenda centered on AI-prompting as anticipatory practice, participatory prototyping with generative AI, and ‘inventive governance’ that resists the self-fulfilling dynamics of predictive systems