Xianglong Zeng

AI systems are treated or even deliberately developed as subjects rather than merely as tools, raising a prospective question of AI subjecthood: under what conditions could such deliberate development genuinely succeed? This article argues that AI subjecthood is a form of normative constitution in which some considerations have non-instrumental normative significance relative to the AI, and such significance is genuinely operative within the AI’s normative organization. Subjecthood in this sense is distinct from consciousness, agency, autonomy, subjectification, welfare subjecthood, and moral status, although these may matter to particular forms of subjecthood. Artificial origin does not by itself preclude AI subjecthood, and the thin conception leaves open what constitutes the relevant artificial bearer, which considerations genuinely have non-instrumental normative significance relative to that bearer, and how such significance becomes operative. In artificial systems, human design and practice inevitably participate in concretizing these dimensions, intentionally or unintentionally; such intervention may therefore be constitutively efficacious without being normatively sovereign: it can help bring about the conditions of subjecthood without thereby reducing artificial subjecthood to human projection, but cannot establish subjecthood by stipulation or rule it out by mere denial. In addition to clarifying the core conditions for genuine artificial subjecthood, the analysis provides an AI-centered basis for ethical analysis, adds an AI-side normative dimension to psychological research on human–AI interaction, frames artificial subjecthood as a multidimensional, cross-cutting engineering concern whose relevance extends to choices that may otherwise appear merely technical, and raises governance questions about institutional control and access to subjecthood-relevant evidence.