Florian Funer¹, Christin Hempeler²

¹ Institute for the Ethics and History of Medicine, University of Tübingen

² Institute of History and Ethics of Medicine, Centre of Excellence in Health Sciences, University Medical Centre Halle, School of Medicine, Martin Luther University Halle-Wittenberg

Deputy treatment decisions are part of everyday clinical practice, particularly in situations where patients have lost their capacity to consent. Advance care planning (ACP) and living wills are intended to provide guidance in such situations, but have not yet become sufficiently widespread. Against this background, there is growing debate as to whether AI-based methods for predicting patient preferences (‘Patient Preference Predictors’) could represent an ethically acceptable alternative.

This article takes this development as an opportunity to conduct a nuanced ethical analysis of the role of AI-based preference prediction in deputy decision-making. It argues that the ethical legitimacy of surrogate decisions does not depend primarily on how accurately presumed preferences are predicted, but rather on whether decisions are based on the reflected, self-expressed will of the person concerned. Even highly accurate algorithmic predictions must still be distinguished, epistemically and normatively, from expressions of will: they reconstruct probabilities, not authorship. Preference prediction thus risks narrowing the meaning of autonomy to the correspondence of decisions with statistically derived preferences and undermining the personal dimension of prior decision-making.

At the same time, it is emphasised that digital technologies and AI have considerable potential to support, rather than replace, ACP processes – for example, through structured support for discussions, low-threshold documentation or improved accessibility. The paper therefore advocates a clear normative prioritisation: AI-supported preference prediction may, at best, serve as a secondary aid in situations of complete disorientation, but must not take the place of ACP and explicit advance directives. For clinical practice, this means consistently aligning technological innovation with the aim of strengthening the prior formation of will.