Can AI Break the Invisible Labor Trap Holding Women Leaders Back?

The AI boom is usually measured in speed: faster drafts, faster research, faster summaries, faster decisions. However, a new paper by Marcela Kostihova and Irina Makarevitch in Merits asks a sharper question: what if AI's real significance lies not in making work faster, but in revealing how much invisible work modern leadership already depends on?

Professional success relies on a dense infrastructure of support: scheduling, drafting, coordinating, remembering, mentoring, smoothing relationships, preparing documents, tracking obligations, and managing emotional expectations. Much of this labor has historically been feminized, undervalued, and treated as background work rather than leadership work.

The authors place generative AI within this longer history of invisible labor. Drawing on Judy Brady's 1971 paper "I Want a Wife" and Ira Levin's The Stepford Wives, they argue that women have long been expected to provide the hidden scaffolding that enables others to succeed. The contemporary AI assistant, they suggest, risks becoming a digital version of that old arrangement unless institutions use it deliberately, transparently, and fairly.

AI can take on routine support work that drains leadership capacity

Generative AI can help with many structured, repetitive, and time-consuming tasks that consume professional attention but rarely define strategic leadership. These include drafting first versions of reports, summarizing meetings, organizing information, synthesizing documents, preparing communications, translating or reformatting content, and generating structured feedback.

Much of leadership is constrained not by lack of intelligence, but by cognitive overload. Leaders are expected to remain responsive, visible, polished, relationally attentive, and continuously productive. For women, the burden is often heavier. The paper points to research showing that women continue to perform disproportionate unpaid household work, workplace service tasks, mentoring, committee labor, and emotional coordination. These responsibilities are essential, but they are rarely rewarded proportionately.

Generative AI could reduce some of this pressure by taking on what the authors call the early-stage or "initiation" work that comes before judgment: the blank page, the first summary, the first structure, the first synthesis. Used responsibly, AI can free human attention for higher-value tasks: strategic thinking, ethical judgment, negotiation, mentoring, public engagement, and decision-making.

For developing countries and Global South institutions, the implications are significant. Women leaders in public administration, education, health systems, NGOs, small enterprises, and community organizations often operate with limited staff and heavy documentation burdens. AI tools could help reduce administrative overload, improve communication, support multilingual work, and expand access to professional-quality drafting and analysis.

However, the benefits depend on affordable access, digital literacy, reliable infrastructure, local-language capability, data protection, and organizational trust. Without these, AI may widen the gap between well-connected professionals and workers excluded from digital transformation.

AI could reproduce same gendered hierarchy it appears to disrupt

Digital assistants have often been feminized in voice, design, and social expectation. Users are encouraged to issue commands to compliant, always-available systems that anticipate needs, smooth friction, and never push back. This design logic echoes older patterns of gendered service.

Technology does not enter an empty workplace. It enters organizations already shaped by unequal expectations. Women may be judged more harshly for using AI assistance than men. A male executive using AI may be seen as efficient and strategic. A woman leader using the same tool may be perceived as less capable, less original, or less personally attentive. The same act of delegation can carry different reputational costs.

The authors also highlight a crucial labor-market trade-off. AI may reduce invisible labor for some women in leadership while increasing insecurity for others. Administrative and clerical jobs, many of them held by women, are among the roles most exposed to automation. A gender-aware AI strategy cannot celebrate productivity gains for senior professionals while ignoring risks for support workers.

In many economies, clerical, coordination, and administrative roles provide stable employment pathways for women. If AI adoption cuts these jobs without reskilling, redeployment, or social protection, the technology could deepen gendered economic vulnerability.

Expectation creep is another risk. If AI makes it faster to produce reports, emails, summaries, and presentations, organizations may simply demand more output. Productivity tools can intensify work rather than reduce it. In that scenario, AI does not liberate workers from invisible labor; it accelerates the pace at which invisible labor is expected.

The policy challenge

Generative AI should not be treated as a private shortcut or secret advantage. Organizations need explicit rules about where AI assistance is appropriate, how outputs should be reviewed, what tasks should remain human-led, and how AI use will be evaluated.

AI policy should include labor-market safeguards, gender-impact assessments, worker protections, digital-skills programs, and accountability standards. AI strategies should not focus only on innovation and competitiveness; they should also address job quality, inclusion, bias, and the distribution of productivity gains.

Employers should identify routine tasks suitable for AI support, train workers to supervise AI outputs, and ensure that AI-assisted work is not stigmatized. They should also monitor whether AI changes the allocation of non-promotable tasks such as note-taking, internal coordination, mentoring, and committee work.

For international organizations and development agencies, the paper points toward a more inclusive digital transformation agenda. AI capacity-building should target women professionals, women-led enterprises, public-sector workers, educators, health administrators, and civil society leaders. However, it should also support workers whose roles may be disrupted by automation.

As for researchers, the paper notes that more empirical work is needed on whether AI actually reduces workload, who benefits most, whether women face an AI-use penalty, and how AI affects promotion, burnout, job security, and workplace evaluation. More research is also needed outside Western professional settings, particularly in multilingual, low-resource, and public-sector environments.

It is important to mention that the paper is not an empirical study. It does not present original survey data, field experiments, or workplace measurements. Its contribution is conceptual and analytical. The paper shifts the AI debate from tools to power. It asks who has historically received support, who has provided it, who is allowed to delegate, and who pays the price when labor becomes invisible.