Philosophy-informed machine learning

Philosophy-informed machine learning (PhIML) directly infuses core ideas from analytic philosophy into ML model architectures, objectives, and evaluation protocols. Therefore, PhIML promises new capabilities through models that respect philosophical concepts and values by design. From this lens, this paper reviews conceptual foundations to demonstrate philosophical gains and alignment. In addition, we present case studies on how ML users/designers can adopt PhIML as an agnostic post-hoc tool or intrinsically build it into ML model architectures. Across all three case studies, PhIML delivered consistent gains with logical and counterfactual violations drastically reduced with minimal accuracy trade-offs (−0.6 to +1.7 %age points (pp)). In one specific fairness setting, worst-off hiring rates rose by more than 23 % and equity gaps reversed, while overall selection increased by + 10.8 pp and ≥ 90 % of baseline accuracy was retained. Finally, this paper sheds light on open technical barriers alongside philosophical, practical, and governance challenges and outlines a research roadmap toward safe, philosophy-aware, and ethically responsible PhIML. • Philosophy-informed machine learning (PhIML) directly infuses philosophy into ML model architectures. • PhIML promises new capabilities through models that respect philosophical concepts and values by design. • We demonstrate philosophical gains and alignment. • We present case studies on how ML users/designers can adopt PhIML as an agnostic post-hoc tool or intrinsically build it into ML model architectures.