Abstract: Ensuring that intelligent machines align with human values and ethical principles is a critical challenge of our time. This paper proposes a novel approach to improve rule-based ethical AI systems. We introduce three principles derived from Martin Heidegger’s philosophy: Autotelism, Sociality, and Finitude as context-sensitive ethical guidelines. These principles ensure that AI prioritizes its proper functioning while recognizing its embeddedness in social and finite contexts. Unlike rigid rule-based systems, which struggle to apply rules in complex and novel situations, our principle-based approach enables adaptive decision-making through a method we term principle-guided situational understanding (PGSU). By integrating these principles, we provide a philosophical foundation that complements technical efforts in AI alignment, fostering machines that are both intelligent and ethically attuned to the human lifeworld.

Keywords: AI Ethics, Ethics, AI Alignment, Heidegger, Heideggerian AI

by K Kraatz & S. Xie