This paper introduces the Ethics-by-World-Notions framework as an approach toward transparent evaluative cognition in artificial intelligence systems. Contemporary AI architectures primarily rely on semantic prediction, statistical approximation, and alignment-based regulation without explicitly representing the evaluative structures underlying normative behavior. The framework developed here argues that future evaluative AI systems require explicit World Notions through which propositions, actions, goals, and outputs become comparatively assessable relative to contextual orientation, obligations, and value systems. Building on the 5-S Model, the paper distinguishes systematically between semantic systems operating through World Models, evaluative systems operating through World Notions, and conscious systems possessing World Experience. The framework further differentiates between implicit evaluative systems and explicit evaluative systems in order to describe the transition from implicit evaluative orientation toward explicit evaluative reasoning and transparent evaluative architectures. The paper additionally argues that classical binary conceptions of truth remain insufficient for describing evaluative cognition because evaluative systems must also process contextual interpretation, normative conflict, intentional distortion, and structured evaluative deviation. Evaluative AI systems therefore require explicit evaluative architectures extending beyond alignment mechanisms, behavioral restriction, and statistical approximation alone. The framework concludes that transparent normative reasoning within future AI systems depends on explicit evaluative structures through which transparent evaluative cognition becomes structurally representable, contextually configurable, and operationally interpretable.
Dirk Gamboa (Mon,) studied this question.