Shoshana Zuboff

In the AI Story

Zuboff's intellectual formation was unusual for a business school scholar. She combined deep ethnographic fieldwork—spending years embedded in workplaces undergoing technological transformation—with theoretical sophistication drawn from sociology, philosophy, and critical theory. Her method was observational rather than prescriptive: she watched what happened when real workers encountered new technologies in real organizational settings, documenting their experiences with a granularity that business scholarship rarely attempted. This grounded empiricism gave her frameworks unusual durability. When In the Age of the Smart Machine was published in 1988, its observations about computerization were specific to the 1980s workplace. Yet the concepts proved generalizable: every subsequent technological transition—from personal computers to the internet to mobile devices to AI—has exhibited the automating/informating dynamic she first mapped in paper mills.

Her transition from studying workplace computerization to analyzing surveillance capitalism reflected her recognition that the mechanisms she had documented in factories and offices had scaled to encompass the entirety of human experience mediated by digital systems. The behavioral surplus she identified in 2019 was not a new phenomenon but the logical extension of the informating dynamic she had studied for three decades. Every technology that generates data about the process it mediates creates the possibility of new knowledge. The question is who builds that knowledge, who owns it, and whose purposes it serves. In the paper mills, the data generated by computerization could have been used to deepen workers' understanding of the production process. In practice, it was used primarily to enable managerial oversight and cost reduction. In surveillance capitalism, the data generated by digital interaction could, in principle, serve the user's interests. In practice, it is extracted, processed, and sold to serve commercial interests that may contradict the user's autonomy.

By December 2025, as the AI transition accelerated, Zuboff's position had become more radical. In interviews and public appearances, she insisted that regulation was insufficient—that the fundamental mechanisms of surveillance capitalism, including the secret massive-scale extraction of human experience and its declaration as corporate asset, must be abolished. This hardening reflected her assessment that the extraction had expanded beyond search and social media into the domain of cognitive labor itself. When users interact with large language models, they generate behavioral surplus that is more intimate and more consequential than any previous form of digital data—data about how they think, create, judge, and reason. The platforms that capture this data possess a form of knowledge about users that could be deployed in hiring, evaluation, and the cognitive sorting of workers by capability. The panoptic sort she had analyzed in Oscar Gandy's terms had evolved into something more comprehensive and more consequential than either she or Gandy had anticipated.

Origin

Zuboff became one of the first tenured women on Harvard Business School's faculty and spent her career there until her retirement, maintaining unusual independence from corporate consulting relationships that shaped many of her peers' research agendas. This independence allowed her to conduct long-term ethnographic studies without the commercial pressures that might have abbreviated or redirected her observations. Her fieldwork in paper mills—spending weeks at a time observing workers on the floor and in newly computerized control rooms—was methodologically closer to anthropology than to business scholarship, and it produced findings that challenged the technology industry's triumphalist narratives about computerization as unqualified progress.

Her intellectual trajectory from workplace studies to political economy to the demand for institutional abolition represents one scholar's sustained attempt to trace a single mechanism—the extraction of human knowledge and experience by technological systems—across four decades of accelerating deployment. The consistency of the pattern she documented, from 1980s paper mills to 2020s AI platforms, gives her framework diagnostic power. The pattern recurs because the economic logic recurs: when technology creates the possibility of replacing human labor while simultaneously generating data about human processes, the default institutional choice is to capture the cost savings and monetize the data, rather than invest in the human capabilities that the data could support. The default is extraction. The alternative requires deliberate, sustained, often politically contested institutional construction. That construction is the work Zuboff has called for since 1988, and its absence is the unfinished question that her entire career has been organized around documenting.

Key Ideas

Automating vs. Informating. The dual, simultaneous dynamics of every smart machine transition—automation replaces human labor; informating generates new data about processes—and the institutional choice between capturing cost savings or investing in new knowledge determines the human outcome.

Action-Centered and Intellective Skill. The embodied, tactile knowledge built through physical engagement with work (action-centered) versus the symbolic, analytical capacity to work with abstracted representations (intellective)—and the dilemma workers face when technology destroys the first while demanding the second.

Behavioral Surplus and Surveillance Capitalism. Human experience claimed as free raw material by digital platforms, processed through machine intelligence into prediction products sold in behavioral futures markets—a system of extraction operating beneath the user's awareness.

The Informating Dividend. The genuine potential expansion of human knowledge and capability that smart machines create—unrealized in most transitions because institutions choose automation over the investment required to capture the dividend.

Epistemic Inequality. The asymmetry between those who possess knowledge about systems and those who are known by systems—amplified by AI to include cognitive sorting based on interaction patterns that reveal thinking itself.

Debates & Critiques

Zuboff's surveillance capitalism framework has been contested by scholars like Cory Doctorow, who argue that behavioral modification claims are overstated and that the real problem is monopoly power rather than epistemic extraction. Her call for abolition rather than regulation has been criticized as politically unworkable. Her framing of AI as continuous with surveillance capitalism has been challenged by researchers who see AI as qualitatively different from advertising-based extraction. The debates are substantive and unresolved, but her empirical documentation of extraction mechanisms operating beneath users' awareness has proven difficult to dismiss.