Abstract This paper proposes the concept of relational AI as a model of longitudinal cognitive partnership between human and artificial intelligence. It argues that, in stable and progressively refined contexts of use, AI may move beyond episodic instrumental use and become part of a functional process of cognitive collaboration. This model is characterized by interactional continuity, iterative calibration, contextual accumulation, shared working language, and the explicit preservation of human responsibility. The paper distinguishes relational AI from both instrumental reduction and excessive anthropomorphism, and discusses its relevance in medicine, research, education, business, and knowledge management.
Luiz Fernando Morrone (2026) studied this question.