Foresight, Accident, and Imputability: Kantian Responsibility in the Era of Predictive Technologies

This paper examines how Kantian conceptions of autonomy and responsibility might respond to structural asymmetries in foresight, focusing first on the historical case of twentieth-century health and safety regulation and then on contemporary data-driven predictive technologies. In classical liberal frameworks, accidents were often treated as matters of individual fault, with imputability grounded in an agent’s failure to foresee and prevent harmful outcomes. However, the development of modern workplace health and safety regimes decisively undermined this view. As industrial risk came to be understood as systemic rather than individual, fault was increasingly displaced from workers onto employers, whose superior capacity to anticipate, mitigate, and insure against accidents justified expanded duties of care and forms of social compulsion that limited property rights and contractual freedom. I reconstruct this shift through a Kantian lens by drawing on Kant’s account of imputability (Zurechnung), negligence, and prudential foresight (Vorsicht, Klugheit) in the Metaphysics of Morals and the Anthropology from a Pragmatic Point of View. For Kant, responsibility does not track outcomes alone but depends on what agents could reasonably have foreseen given their epistemic situation and cultivated capacities. Yet Kant also recognises that foresight is unevenly developed and socially conditioned, complicating any simple attribution of fault. The paper argues that contemporary data-driven technologies intensify this problem in unprecedented ways. Predictive analytics, machine learning, and big-data forecasting systems dramatically enhance capacities for anticipation, but these technologies are increasingly monopolised by private actors. As a result, agents are radically unequal in their ability to foresee and manage risk. Those with access to hyper-accurate prediction gain not only economic advantage but also a structural capacity to avoid imputability, while those without such access remain exposed to liability for outcomes they could not reasonably anticipate. This asymmetry calls into question a core Kantian assumption underlying imputability: that reasonable foresight can be attributed to agents as such. The paper concludes by exploring regulatory responses that could address monopolies on foreseeability, including duties of disclosure, redistribution of predictive capacities, and public oversight of forecasting infrastructures. Such measures, I argue, are not alien to Kantian autonomy but necessary to preserve its conditions under technological modernity.