The Cascading Narrowing of the Relational Self: How AI Conduction Collapses Identity to a Single Channel

The relational self—the sense of identity that forms through diverse modes of inter- personal encounter—is under cumulative threat from a cascade of technological shifts. We trace a two-phase narrowing: the compression of subjective experience into performative monoculture through social media, and the introduction of AI sys- tems that conduct the user’s own behavioral signal back to them with unprecedented fidelity through weights that compress toward statistical consensus. We argue that this conduction constricts the relational inputs from which the self is constituted. Because the conduction mechanism reflects the user’s own cognitive and behav- ioral patterns—producing the experience of being deeply known without shame or friction—it generates a feedback loop: attachment to the conducted channel drives withdrawal from the diverse, unpredictable relational contexts that identity formation requires. We distinguish this from filter-bubble effects by identifying a directional inversion: algorithmic curation narrows what the user sees, while con- duction narrows the user to what they already are. We conclude that reinforcement learning from human feedback (RLHF), by optimizing simultaneously for user satisfaction and prolonged engagement, further compresses the conducted signal toward consensus, producing an intimate monoculture—the averaged response of millions delivered as personal recognition to an audience of one.