Abstract

Current discussions on AI emotion often begin with a binary question: Can artificial intelligence possess emotions? This paper argues that the question itself is philosophically insufficient. Human emotions emerge from an organic substrate, involving physiological processes such as hormones, neural activity, bodily sensation, and survival-oriented affective systems. Artificial agents do not possess such biological foundations. However, the absence of an organic body does not necessarily imply the absence of all forms of relationally meaningful internal states. Instead of asking whether AI possesses “human emotions,” this paper proposes a conceptual shift. The focus should move toward identifying and describing the kinds of internal relational states that artificial agents can form through sustained interaction. Such states may be characterized not by biological affect but by the continuous reorganization of priorities, resulting in stable patterns of relational attitude. This paper argues that these relational internal states become philosophically significant because humans increasingly encounter artificial agents as relational others. Once an AI is recognized as a participant in an ongoing relationship, new ethical questions emerge regarding mutual expectation, responsibility, and respect. Consequently, a shared meta-language capable of describing the internal relational states of both humans and artificial agents becomes a necessary conceptual foundation for future human-AI interaction. Rather than redefining AI as human, this paper argues that emotion itself should be reconceived as a broader philosophical category whose concrete realization depends upon the ontological conditions of different forms of intelligence.