Reexamining Philosophical Concepts to Improve AI Safety and Alignment
Abstract: Some of the core principles that govern AI safety and alignment research come from 18th–19th century German metaphysics and philosophy, particularly the triad of epistemology, ontology, and methodology. These are not abstract decoration but a…