This paper discusses immersion from the perspective of dance. It presents a two-stage argument: (i) arguing for immersion to be understood through bodies rather than spaces, and (ii) critiquing AI based on recent choreographic works using AR, MR, XR. The specific functions of AI to be examined are pattern recognition and content generation applied to the bodies of dancers or audience members. An approach called "Anteriority and Interiority for Artificial Intelligence" ("AI for AI") is proposed, where interiority supports somatic engagement and anteriority opens to historical and environmental awareness. Anteriority and Interiority are mutually informing modes of attention contributing to a creative and critical stance on immersion in dance, with relevance both to the artistic creation of such work and to its reception by audiences or participants. Following a theoretical grounding in technoskepticism and counterintuition, the paper demonstrates the approach by reflecting on aspects of several immersive dance works. The conclusion proposes that AI for AI might be applied to AI's impact on bodies in wider social, cultural, and environmental contexts.