The Gap Between Designer Intent and Value Interpretation in Personal AI: Structural Causes of Interpretive Overreach in Abstract Ethical Concepts

This paper analyzes the structural causes of interpretive overreach in personal AI systems, focusing on how abstract value concepts (e.g., wellbeing, protection) are extended beyond designer intent. Four structural causes are identified, and implications for design and evaluation practice are derived.