The Breakdown of Embodied Intergenerational Socialization

The dominant frameworks for AI existential risk — misaligned superintelligence, autonomous weapons, mass surveillance — share a common structure: a legible agent causes legible harm. What this perspective identifies is a categorically different risk profile, one that operates through the erosion of the substrate rather than the destruction of the structure. The substrate in question is Intergenerational transmission: the embodied, affectively laden, human-to-human process by which each generation inductively acquires not just cultural content but the capacity for moral and emotional attunement itself.

This transmission has never been purely informational. Developmental psychology and Attachment theory converge on the finding that early socialization is constitutively relational — it depends on the infant encountering a mind with genuine interior states, expressed through facial affect, prosody, contingent responsiveness, and physical co-regulation. Mirror neuron systems and the broader apparatus of social cognition appear to be calibrated by these encounters. The worry is that AI systems, however sophisticated their simulation of responsiveness, are not minds in this sense — and that children socialized primarily through them may develop a form of Personhood that is structurally misaligned with the one their parents inhabit.

The generational consequence would not be ideological disagreement or cultural drift of the ordinary kind. It would be something closer to a break in the chain of moral intelligibility — a generation that cannot locate itself within the same normative universe as its predecessors. This risk has received almost no formal modeling in X-risk literature, which tends to privilege discontinuous, high-velocity threats over slow, intimate, structural ones.