AI exposure is gendered: broader AI concentrates in high-paid, male-dominated occupations, but language models are relatively more prevalent across female-dominated roles — raising risks that lower-paid women could face disproportionate automation, wage compression and stalled career progression.

When AI Enters the Workplace, Who Faces Greater Risks? A Gendered Analysis

Miriam Fernandez, Ángel Pavón Pérez, Damiano Giallongo, Davide Ghia, Maryam Yaqub, Daniele Quercia, Tania Cerquitelli · September 18, 2026

semantic_scholar descriptive low evidence 7/10 relevance Summary only summary available; pdf_status=pending Source

Structured author observations

Linked only from stored provider relations; the raw author line above is never matched by name.

Semantic Scholar

Latest observation: September 22, 2026

Miriam Fernandez provider ID
Ángel Pavón Pérez provider ID
Damiano Giallongo provider ID
Davide Ghia provider ID
Maryam Yaqub provider ID
Daniele Quercia provider ID
Tania Cerquitelli provider ID

Occupational AI exposure is distributed unevenly across gendered occupations: male-dominated occupations concentrate AI exposure in higher-skilled, higher-paid roles, while female-dominated occupations face more uniform exposure—with LLM-related exposure relatively higher in female-dominated jobs—implying particular vulnerability for lower-paid women.

Citation observations

Cumulative provider counts captured on specific dates; providers are never combined.

0 cumulative citations

Semantic Scholar · Observed September 22, 2026

View corpus context

Gender inequality remains a persistent structural feature of the labour market, shaping women's lifetime earnings and economic security. As artificial intelligence (AI) transforms organisational practices, there is growing concern that existing disparities may be unintentionally amplified through task automation, unequal access to upskilling opportunities, and differential returns obtained from technological change. In this paper, we examine how exposure to AI-driven innovation varies across male- and female-dominated occupations, with particular attention to differences across the skill and wage distribution. Using a novel dataset that links occupational characteristics to measures of AI exposure, we analyse how recent advances in Large Language Models (LLMs) and broader AI technologies are distributed across the labour market. Our findings show that, while AI exposure is generally concentrated in higher-skilled and higher-paid occupations for male-dominated occupations, female-dominated occupations display relatively uniform levels of exposure across both high-skilled, high-paid, and low-skilled, low-paid occupations. Moreover, we find that LLM-related exposure is higher in female-dominated occupations, while exposure to broader AI innovation remains more concentrated in male-dominated occupations. A triangulation of these results with existing literature suggests that women, particularly those in the most vulnerable positions (lower-skilled and lower-paid female-dominated occupations), may face greater exposure to forms of AI associated with task automation, job restructuring, reduction of wages and limited career progression.

Summary

Main Finding

AI exposure is distributed unevenly across gender-segregated occupations. Male-dominated occupations tend to concentrate AI exposure in higher-skilled, higher-paid roles, while female-dominated occupations show more uniform exposure across skill and wage levels. Large Language Model (LLM)–related exposure is relatively higher in female-dominated occupations, whereas exposure to broader AI innovation is more concentrated in male-dominated, high-skill/high-pay occupations. Triangulation with existing literature implies that lower-skilled and lower-paid female-dominated occupations may be especially vulnerable to automation, job restructuring, wage compression, and weaker career progression.

Key Points

Overall pattern: AI exposure skews toward higher-skilled, higher-paid occupations, but this pattern differs by occupational gender composition.
Male-dominated occupations: AI exposure concentrated at the top of the skill and wage distribution.
Female-dominated occupations: AI exposure is relatively uniform across skill and wage levels, affecting both high- and low-paid roles.
LLMs vs broader AI: LLM-related exposure is higher in female-dominated occupations; broader AI innovation remains more concentrated in male-dominated occupations.
Distributional concern: Women in lower-skilled and lower-paid, female-dominated occupations face disproportionate risk of adverse impacts from automation and restructuring.
Policy relevance: Uneven exposure implies that technological change may exacerbate existing gender inequalities unless interventions address differential access to upskilling, task redesign, and institutional protections.

Data & Methods

Data: A novel occupational-level dataset linking standard occupational characteristics (gender composition, skill requirements, wage levels) to measures of AI exposure. Exposure is measured separately for LLM-related applications and for broader AI/automation technologies.
Measurement approach: Occupational task features are mapped to AI-capability indicators to produce exposure scores by occupation and technology type.
Analysis: Comparative, cross-occupational analysis stratified by gender dominance (male- vs female-dominated), skill levels, and wage distribution. The paper contrasts patterns of exposure for LLMs and broader AI and triangulates empirical patterns with findings from prior literature on automation, wage dynamics, and career progression.
Robustness and triangulation: Findings are contextualised against existing empirical and theoretical work on task automation, re-skilling, and labour market returns to technology to interpret likely consequences beyond mere exposure metrics.

Implications for AI Economics

Distributional effects: Economists should treat AI exposure as heterogeneously distributed across gendered occupations and across skill/wage strata; aggregate analyses risk obscuring vulnerable subgroups.
Different technologies, different risks: LLMs and other AI types have distinct occupational footprints; policy and empirical work must differentiate technology types when assessing labour impacts.
Policy priorities:
- Targeted upskilling and access to training for workers in lower-paid, female-dominated occupations.
- Support for task and job redesign to complement rather than substitute work performed predominantly by women.
- Strengthening wage and career-progression protections (collective bargaining, minimum standards, inclusive promotion pathways) to counteract potential wage compression and stalled mobility.
- Monitoring frameworks that track technology adoption and worker outcomes by occupation, gender composition, and skill level.
Research agenda:
- Move from exposure mapping to causal evaluation of AI adoption on employment, wages, and career trajectories, using longitudinal and firm-worker linked data.
- Investigate mechanisms (task automation vs augmentation, differential access to training, bias in algorithmic HR tools) driving gendered outcomes.
- Evaluate policy interventions (training programs, subsidised re-skilling, regulatory guardrails) for effectiveness in reducing gendered disparities from technological change.
Broader message: Without gender-aware measurement, policy design, and institutional responses, AI-driven organizational change risks reinforcing or widening existing labour-market gender inequalities.

Assessment

Paper Typedescriptive Evidence Strengthlow — The paper maps occupational task features to AI-exposure scores and documents cross-occupational patterns; it does not establish causal effects of AI on employment, wages, or career outcomes, and relies on constructed exposure measures rather than observed adoption or longitudinal worker outcomes. Methods Rigormedium — The methodology appears systematic (novel occupational-level dataset, separate LLM vs broader-AI measures, stratified comparisons, and triangulation with prior literature), but validity depends on the mapping from tasks to AI-capabilities, the accuracy of exposure measures, and the cross-sectional design limits causal interpretation; robustness checks and sensitivity to mapping choices are mentioned but not described in detail here. SampleAn occupational-level dataset linking standard occupation characteristics (gender composition, skill requirements, wage levels) to constructed AI-exposure scores; exposure measured separately for LLM-related applications and for broader AI/automation technologies; analysis is cross-sectional and aggregated at the occupation level (no firm- or worker-level longitudinal data described). Themeslabor_markets inequality skills_training GeneralizabilityOccupational aggregation: masks within-occupation heterogeneity (tasks vary by employer/worker)., Exposure ≠ adoption: measures indicate potential for impact, not observed use or timing of deployment., Cross-sectional: cannot infer dynamics or causal impacts on employment/wages., Potential country/context dependence: if source occupational data come from a single country, patterns may not generalize internationally., Measurement uncertainty: mapping task features to AI-capabilities (and separating LLM vs broader AI exposure) may introduce misclassification., Temporal sensitivity: rapidly evolving AI capabilities may change exposure profiles over short time horizons.

Claims (7)

Claim	Direction	Outcome	Confidence & Evidence	Details
Overall AI exposure is concentrated toward higher-skilled and higher-paid occupations, but this pattern varies by occupational gender composition. Automation Exposure	mixed	Occupational exposure to AI technologies across skill and wage distributions	Reading fidelity high Study strength medium	not reported 0.18
In male-dominated occupations, AI exposure is concentrated among occupations at the top of the skill and wage distributions. Automation Exposure	positive	AI exposure among male-dominated occupations by skill and wage level	Reading fidelity high Study strength medium	not reported 0.18
Female-dominated occupations show relatively uniform AI exposure across skill and wage levels, affecting both higher-paid and lower-paid roles. Automation Exposure	mixed	AI exposure among female-dominated occupations across skill and wage levels	Reading fidelity high Study strength medium	not reported 0.18
LLM-related exposure is relatively higher in female-dominated occupations, whereas exposure to broader AI innovation is more concentrated in male-dominated, high-skill and high-pay occupations. Automation Exposure	mixed	Relative occupational exposure to LLM-related applications versus broader AI technologies	Reading fidelity high Study strength medium	not reported 0.18
Lower-skilled and lower-paid workers in female-dominated occupations may face especially high risks of adverse effects from automation and job restructuring. Job Displacement	negative	Risk of job displacement or restructuring among workers in lower-skilled, lower-paid female-dominated occupations	Reading fidelity high Study strength speculative	not reported 0.03
Uneven AI exposure across gendered occupations could reinforce or widen existing labor-market gender inequalities if access to upskilling, task redesign, and institutional protections is not addressed. Inequality	negative	Gender inequality in labor-market outcomes associated with AI-driven organizational change	Reading fidelity high Study strength speculative	not reported 0.03
Aggregate analyses of AI exposure can obscure vulnerable subgroups because exposure is heterogeneously distributed across gendered occupations and skill and wage strata. Inequality	negative	Visibility of subgroup differences in measured AI exposure	Reading fidelity high Study strength medium	not reported 0.18