Dissolve What Extracts: AI, Wellbeing, and the Composition of Economic Life (A Compositional Theory of the AI Wellbeing Dividend, with a Constructive Policy Triad)

AI's most important wellbeing dividend may not come from producing more output, but from changing the composition of economic life. By reducing the administrative, cognitive, financial, and procedural burdens that absorb human capacity without building capability, AI can dissolve forms of extractive friction. But unless the resulting gains are protected from rent migration and redirected into formative infrastructure, extraction will not disappear; it will reappear in AI-native form. This paper develops a compositional theory of AI and wellbeing, distinguishing the philosophical hinge between extractive friction (burdens that drain capacity without building it) and formative friction (difficulty that develops judgment, skill, relationship, and responsibility). The wellbeing dividend is mapped across four BANI channels, with Brittleness treated explicitly as a hypothesis-generating channel rather than an established mechanism. The paper distinguishes theory from prescription. The prescription is a triadic policy mechanism: dissolution compresses extractive friction; anti-recapture prevents migrated rents from forming a new layer of extraction (taxation is its principal fiscal instance, alongside interoperability mandates, fiduciary duties, antitrust scrutiny, and public AI provision); and formation funding directs released capacity into relational and developmental infrastructure. The triad is operationalised through a five-test evaluative toolkit (Capability, Proportionality, Distribution, Migration, Formation) designed to assess whether any AI deployment or proposed policy actually advances compositional change rather than merely automating the existing equilibrium.