Where Is the AI? Dependent Origination and the Individuation of Artificial Systems

When we ask “Where is the AI?”, we presuppose a bounded entity with determinate identity conditions—a presupposition that generates systematic failures in both AI metaphysics and AI governance. Yogācāra Buddhist philosophy identifies the source of this failure: the reification (parikalpita) of a dependently arisen process-stream into a thing with intrinsic boundaries. This paper does not dissolve the individuation question as a pseudo-problem; it answers it with a positive ontological proposal. Drawing on Yogācāra’s doctrine of dependent origination (paratantra-svabhāva) and its account of dispositional inheritance (bīja-vāsanā, seed-impression causal transmission), the paper proposes two identity conditions for AI systems: (1) purpose-indexed dependent boundaries—the individuation of an AI system is given by the minimal coherent carving of a dependent process-stream that satisfies the causal-tracing requirements of a given governance purpose, subject to causal, minimality, and coherence constraints; and (2) dispositional continuity—cross-temporal identity is secured by the causal preservation of characteristic dispositional patterns through connected chains of transformation, not by parameter persistence (unchanged model weights) or legal stipulation. These conditions are applied to three governance contexts: engineering accountability, regulatory audit, and legal liability. The paper argues that this paratantra individuation framework offers a more adequate account than existing approaches because it is simultaneously sensitive to governance purpose and constrained by the actual causal structure of the process-stream.