Research Information System
AI and Society, 2026 (ESCI, Scopus)
This paper argues that contemporary AI intensifies a long process of somatic alienation, the externalization of human capacities, organ by organ, to tools, by increasingly targeting judgment and cognitive capacities rather than mere effort. The claim is conditional: under prevailing ownership, deployment, and incentive regimes, many AI systems can push practices beyond supportive relief toward domination; yet with alternate designs and governance, relief can be non-alienating. To distinguish these trajectories, the paper proposes a practical two-out-of-three audit. A technology is provisionally alienating when at least two hold. (i) Opacity, where users cannot predict or contest outputs. (ii) Irreversibility, where skills or infrastructure atrophy such that re-entry is costly. (iii) Displaced ends, where optimization objectives shift from users to institutional or commercial reasoning. Applying this audit across cases, such as in navigation, writing, coding, or diagnosis, reveals movement along a four-mode progression. The pattern runs from augmentation to delegation to integration and, in some deployments, to domination of human ends. The paper makes three original contributions. (1) A four-mode model that unifies physical and cognitive externalization without teleology. (2) A two-out-of-three audit that operationalizes alienation with observable indicators for researchers, designers, and regulators. (3) Design norms that preserve evaluative agency, maintain reversibility, and enforce legibility, for example, user user-settable goals, rationale exposure, and skill retention protocols. Philosophically, I synthesize Anders, Stiegler, Arendt, and Gehlen to clarify when “relief” becomes rule in the historical trajectory.