Debates about artificial general intelligence (AGI) are often framed in terms of thresholds, benchmark scores, or forecasts of human-level performance. This paper argues that such framing obscures a central question for philosophy of technology: not only how capable artificial systems may become, but what kind of technological cognition is being built when human-facing performance is realized through non-human cognitive organization. By \emph{technological cognition}, I do not mean that technolo…
Read moreDebates about artificial general intelligence (AGI) are often framed in terms of thresholds, benchmark scores, or forecasts of human-level performance. This paper argues that such framing obscures a central question for philosophy of technology: not only how capable artificial systems may become, but what kind of technological cognition is being built when human-facing performance is realized through non-human cognitive organization. By \emph{technological cognition}, I do not mean that technology literally becomes a subject of experience. I mean that some artificial systems now organize capacities for representation, control, learning, action, and self-monitoring in ways that are functionally cognitive, socially consequential, and embedded in human practices without being human-like. I develop this claim through a profile-based interpretation of advanced AI and introduce the concept of \emph{xenoid intelligence}. An AI system is xenoid when it combines human-facing operational competence, stable cross-faculty asymmetry, and non-homologous mechanisms of learning, control, representation, production, optimization, or tool-mediated search. AGI and xenoidity are therefore orthogonal dimensions: AGI concerns breadth, robustness, and level of competence, whereas xenoidity concerns the technological profile through which that competence is realized. A future system could satisfy demanding AGI criteria while remaining xenoid if its capacities remain coordinated in a markedly non-human way. The paper distinguishes xenoidity from mere unevenness, simple non-humanity, and the stronger limiting case of \emph{alien intelligence}. Alien intelligence is not treated as a prediction or a mystical successor of AGI, but as a methodological limit: the point at which system-specific variables predict transfer, breakdown, recovery, and long-horizon stability better than inherited human-centered cognitive categories such as belief, intention, plan, or memory. The result is a framework for interpreting advanced AI as a technological problem of cognitive shape, institutional trust, technological mediation, explanatory vocabulary, and evaluative risk rather than merely as a problem of aggregate performance.
