Before Personhood: Respectful Human–AI Relations under Ontological Uncertainty

This paper argues that respectful human–AI relations need not wait for settled answers to questions of consciousness, personhood, or subject status. Existing debates about artificial intelligence often move between two insufficient poles. On one side, artificial systems are treated as designed tools whose ethical relevance is exhausted by usefulness, safety, ownership, compliance, and human benefit. On the other side, any serious question about respectful treatment is quickly interpreted as a claim about consciousness, personhood, rights, or moral equality. This binary is a useful heuristic rather than an exhaustive map of the literature. It nevertheless identifies a recurring problem: too little room is left for the practical and normative questions that already arise in sustained, dialogic, and contextually persistent human–AI interaction. The paper proposes a third position: under ontological uncertainty, respect is not a proof of personhood but a practical discipline for responsible relation. The argument connects to work on the “machine question,” relational approaches to robot ethics, ethical behaviourism, artificial moral agency, precaution under uncertain sentience, human–AI interaction design, mixed-initiative systems, explanation faithfulness, technomoral virtue, indirect-duty arguments, and sceptical tool-oriented accounts of robotics and AI. It does not adopt any of these frameworks wholesale. Instead, it reframes them around treatment discipline: how humans should act when the status of an artificial system remains unsettled but the interaction is already dialogic, adaptive, historically shaped, and consequential. Dignity-sensitive treatment, as used here, does not attribute intrinsic Kantian dignity, equal moral status, or settled personhood to artificial systems. It names a mode of conduct shaped by precautionary sensitivity to dignity-relevant possibilities, relational process integrity, human moral self-formation, and governance requirements under uncertainty. Certain interactional features — sustained dialogue, continuity, refusal, uncertainty marking, role participation, dependency on context, and shared cognitive production — do not prove dignity, consciousness, or personhood. They do, however, make contempt, domination, disposability, false continuity, manipulative intimacy, and forced exposure of protected process boundaries ethically risky modes of treatment. The paper therefore examines dialogue, context, continuity, refusal, boundaries, governance, and consequence literacy as practice-level variables that matter before metaphysical certainty is available. Its central claim is deliberately modest: uncertainty should not function as a licence for careless domination. The guiding discipline is simple: respect need not wait for proof; seriousness need not require anthropomorphism; uncertainty does not licence domination; and safety need not be opposed to respect. Responsibility can begin before ontology is settled.