Most philosophy of artificial intelligence asks whether machines are truly intelligent, whether they understand, or whether they can become conscious. This paper argues that these questions, although legitimate, are not the most historically decisive ones. The deeper event is that intelligence has become artificial: a set of cognitive functions can now be externalized from an organism, represented in technical systems, reproduced at scale, recombined across tools and institutions, revised through design, and delegated into consequential action. Artificiality is therefore not a defect opposed to authenticity. It is a transformation in the ontological, epistemic, and political status of intelligence. The paper develops an artificiality thesis. Intelligence is treated not as an occult substance or a single measurable essence, but as a relational capacity to construct action-relevant representations, manage uncertainty, preserve goals, select among alternatives, and revise through feedback. When portions of this organization are engineered, intelligence becomes an artifact within the world and an object of intelligence itself. This produces a metacognitive reversal: the faculty by which civilization knows the world becomes something civilization can model, manufacture, distribute, and redesign. Yet artificiality also creates a delegation gap. Competence does not imply consciousness; prediction does not amount to judgment; causal efficacy does not confer authority; and successful outcomes do not by themselves establish responsibility. The decisive philosophical problem is therefore not merely whether a machine can think, but how constructed cognition becomes entitled to act. The paper proposes accountable artificiality as the governing ideal: an artificial cognitive act must bind object, information, inference, alternatives, authorization, execution, consequence, evidence, and responsibility into a reconstructable chain. AI matters because it is artificial. It becomes worthy of trust only when its artificiality is made governable.
Xianhao Song (Tue,) studied this question.