The Extraction of Experience

In the AI Story

Zuboff traces the extraction's genealogy through capitalism's history of converting non-commodities into commodities: land (through enclosure), labor (through wage relations), money (through credit systems), and now experience (through computational apparatus). Each conversion required violence—the dispossession of commoners from land, the proletarianization of peasants into workers, the financialization of relationships. The extraction of experience operates through terms-of-service agreements rather than literal violence, but the structural operation is analogous: separating people from a dimension of their lives that was never produced for sale and converting it into proprietary assets generating returns for platform owners. The violence is epistemic rather than physical—users experience extraction as the natural operation of free services rather than recognizing it as the expropriation of cognitive commons.

The AI moment extends extraction into domains surveillance capitalism's earlier architecture did not reach. Search behavior reveals what users want to know; social media reveals what they want to project; AI interaction reveals how they think. When Edo Segal writes The Orange Pill with Claude, every conversation externalizes his creative process: which connections he pursues, which he abandons, how he evaluates suggestions, how long he deliberates before decisions, the rhythm and architecture of his doubt. This cognitive surplus is claimed by Anthropic as raw material for model improvement—used to train systems that will serve millions of users Segal will never meet, generating value he will never share. The extraction is not incidental to the collaboration but structural: the tool's improvement depends on the data the collaboration generates, and the data belongs to the platform by contractual claim users cannot meaningfully contest.

Zuboff's December 2025 El País interview hardened her position from regulation to abolition: "There are very few things left in this world that we can do without contributing to it. That's what makes it intolerable." The statement names the extraction's totalizing character—the progressive elimination of unmonitored, unextracted human activity. By 2026, AI tools had penetrated knowledge work so comprehensively that opting out meant professional marginalization. The choice was not whether to be extracted from but how much to consent to extraction's terms. Zuboff's framework identifies this as the endpoint surveillance capitalism was always approaching: a world where human experience cannot be lived without being claimed, where the capacity to act without generating extractable data has been eliminated, where autonomy itself has become a theoretical possibility rather than a lived condition.

Origin

The concept crystallizes in The Age of Surveillance Capitalism (2019), though its intellectual roots run through Zuboff's entire career. The phrase "extraction of experience" does not appear as such in the text—Zuboff's preferred term is "expropriation of the commons"—but the operation she describes is extraction in the precise sense: the separation of a resource from those who generated it and its conversion into property claimed by others. The framework synthesizes Marx's primitive accumulation, Polanyi's fictitious commodities, and Arendt's analysis of totalitarianism's claim on human life's totality into a diagnosis specific to digital capitalism's computational apparatus.

Key Ideas

Unilateral claiming. Experience is extracted without meaningful consent—terms of service are unreadable, alternatives are unavailable, participation in digital life requires accepting extraction as admission price.

AI as extraction apparatus. Machine intelligence is not separate from surveillance capitalism but its operational mechanism—the factory where behavioral surplus is processed into prediction products whose accuracy depends on extraction's scale and intimacy.

Cognitive surplus is most valuable. AI interaction data reveals professional competence, judgment architecture, creative processes—more commercially valuable than purchase history because it exposes how people think rather than what they want.

Training the replacement is structural. Workers using AI to augment performance simultaneously train systems that will automate their roles—the feedback loop is automatic, unavoidable, built into every interaction's architecture.

Extraction's totalizing trajectory. The progressive elimination of activity that can be conducted without generating extractable data—the endpoint where human autonomy becomes theoretical rather than lived, where opting out means marginalization.