The Eliminated Laundress

In the AI Story

Cowan's documentation of the eliminated laundress drew on census data, city directories, oral histories, and labor statistics showing that domestic service—particularly laundering—was one of the largest occupational categories for Black women in early-twentieth-century American cities. In Atlanta in 1870, roughly five thousand Black laundresses constituted the city's largest single occupational group. By 1960, the occupation had nearly vanished, replaced by household washing machines and the unpaid labor of the women who operated them. The elimination was not immediate—it unfolded across decades—but it was systematic, following the adoption curve of the washing machine with predictable precision.

The economic logic of elimination was individually rational and collectively catastrophic for the affected workers. Each household that purchased a washing machine made a decision that made sense: reduce recurring labor costs by substituting a capital investment for wage payments. Aggregated across millions of households, those individual rational decisions eliminated an entire occupational category, with no social safety net, no retraining programs, no institutional structure to absorb the displaced workers. The laundresses found other work—often as domestic servants in different capacities, sometimes in factories, sometimes in informal economies that left no historical record—but they found it at individual cost that no economic measure counted as a cost of technological transition.

The AI parallel is the data laborer—the content moderator in Manila, the training-data annotator in Nairobi—whose work makes AI possible while remaining invisible to the system's users. Mary L. Gray and Siddharth Suri's Ghost Work documented this global labor force performing microtasks at piece rates that, like the laundress's wages, are meager but real. As AI systems become more capable through advances in self-supervised learning and synthetic data generation, the economic logic that eliminated the laundress is positioning itself to eliminate the data labeler. The work will not disappear—it will be transferred to the machines themselves (through training on AI-generated data) and to the end users (through unpaid evaluation and correction labor). The redistribution will follow the same fault lines: the marginalized workers will lose employment, the privileged users will gain unpaid shadow labor, and the technology will reinforce existing inequalities while appearing to democratize capability.

Origin

Cowan's attention to the laundress came from her broader methodological commitment to centering the experiences of the people who performed the work rather than the people who designed the tools. The laundress was absent from most technology histories, which focused on the inventors and manufacturers of washing machines. Cowan insisted that the laundress was the more important historical figure, because her elimination was the social consequence that the technology's adoption produced—and social consequences, in her framework, matter more than technical capabilities. The figure of the eliminated laundress became a case study in how to write technology history from the perspective of those whom the technology displaced rather than those whom it enriched.

Key Ideas

Technology redistributes labor along existing hierarchies. The washing machine did not randomly distribute its burdens and benefits—it followed the social structure, eliminating paid work for working-class women of color and creating unpaid work for middle-class white women.

Individually rational, collectively catastrophic. Each household's decision to replace the laundress with a machine made economic sense; millions of such decisions eliminated an occupational category without any institutional mechanism to absorb the displaced workers.

The AI data labeler is the contemporary laundress. Performing essential but invisible work at meager wages, positioned for elimination as AI systems require less human-labeled training data, with no safety net and no institutional voice.