Simone Conversano (2026) article proposes the concept of “functional intelligence” as an intermediate category between simple agency and subjective understanding. Developing Floridi’s thesis of the multiple realisability of agency, the author asks whether it is justified to restrict this multiple realisability to agency alone, excluding intelligence. The commentary analyses the criteria of functional integration and the complexity threshold that constitute this definition, pointing to the need to distinguish representation from effective predictive correlation. I also discuss the concept of functional semantics, propose a three-level taxonomy of agency, and indicate the ethical implications of adopting this category for understanding the status of large language models.
