The rapid advance of artificial intelligence is reshaping the global labour market, prompting workers everywhere to ask which occupations will survive the wave of automation. While many assume that jobs centered on human interaction—such as child care, nursing, and aged care—are safe havens, recent research from Australia’s Jobs and Skills Agency reveals a counterintuitive truth: the sectors with the highest concentration of women are actually the most vulnerable to AI‑driven displacement. This insight challenges prevailing stereotypes and forces policymakers, employers, and educators to rethink where retraining and support efforts should be directed. By exposing the hidden gender dimension of AI’s impact, the analysis provides a clearer picture of where the next wave of job transformation will hit hardest.
Conventional wisdom suggests that because women dominate caring professions that rely on empathy and personal touch, their employment is insulated from technological substitution. However, the data shows that the bulk of female‑concentrated work lies not in direct care but in clerical, administrative, and support functions that underpin almost every industry. These roles—ranging from secretaries and receptionists to bookkeepers and payroll clerks—are heavily reliant on repetitive, rule‑based tasks that AI systems can now perform with increasing speed and accuracy. As a result, the very occupations that employ nearly one‑in‑five working women in Australia are showing the strongest automation exposure scores, a metric designed to quantify the susceptibility of job tasks to machine substitution.
The agency’s automation exposure score evaluates each occupation by estimating the proportion of its core activities that could be replicated by existing AI technologies, focusing on tasks that are routine, predictable, and data‑driven. Scores are derived from a detailed task‑level analysis that maps occupational descriptions against current AI capabilities in areas such as natural language processing, pattern recognition, and decision automation. This methodology allows researchers to move beyond vague fears of “robots taking jobs” and instead pinpoint where specific skill sets are most likely to be complemented—or replaced—by algorithmic tools. The resulting scores provide a quantitative foundation for observing patterns across gender, industry, and skill level.
Clerical and administrative occupations consistently rank among the highest in exposure, confirming economists’ expectations that AI excels at streamlining repetitive, rule‑based work. What is striking, however, is the demographic makeup of this high‑risk group: more than 70 % of workers in these fields are women, and the sector accounts for almost 20 % of all female employment nationwide. This concentration means that a significant share of women’s livelihoods is tied to occupations that the data flags as most susceptible to automation. The overlap between high exposure and high female representation is not a statistical fluke; it reveals a structural bias in how technology is being deployed across the economy.
When the researchers examined the twenty occupations with the highest automation exposure scores, a clear gender pattern emerged: fifteen of those roles are female‑dominated, and five of them employ at least 80 % women. Jobs such as secretaries, receptionists, bookkeepers, and accounting, human resources, and payroll clerks all appear in this top‑tier risk list. In stark contrast, only a single occupation among the top twenty—financial dealers—shows a moderate male bias, while the remaining four are roughly gender‑balanced. This imbalance underscores that the AI‑automation wave is not hitting the workforce uniformly; it is propagating along existing occupational segregation lines.
At the opposite end of the spectrum, the twenty occupations with the lowest automation exposure scores tell a different story. Seventeen of these low‑risk roles are male‑dominated, while only three have a notable female concentration. Examples include trades such as concreting, electrical work, plumbing, and construction, where tasks require physical dexterity, real‑time problem solving, and adaptation to unpredictable site conditions. These attributes remain difficult for current AI and robotics to replicate reliably, which helps explain why employment in these areas is holding steady or even growing despite broader automation pressures.
Labor‑market vacancy trends offer a real‑time window into how these exposure differences are already translating into hiring shifts. Data collected by Jobs and Skills Australia over the past three years shows a pronounced decline in advertised openings for many of the high‑exposure, female‑concentrated roles. For instance, vacancies for personal assistants and bookkeepers in June 2026 were roughly 22 % lower than the same month a year earlier, while openings for general clerks fell by about 9 %. These declines far outpace the modest 1.7 % weakening observed across the total workforce vacancy index, suggesting that AI‑related displacement is already exerting measurable pressure on specific job categories.
Conversely, occupations that scored low on automation exposure are experiencing growth in demand, reinforcing the idea that certain skill sets remain uniquely human. Vacancies for concreters increased by approximately 5.5 % over the same period, and electrician openings rose by a robust 14.3 %. These gains reflect ongoing infrastructure projects, renewable‑energy installations, and the persistent need for skilled tradespeople who can interpret complex physical environments, troubleshoot equipment on the fly, and adhere to safety standards that require nuanced judgment—capabilities that current AI systems struggle to emulate consistently.
The persistence of hands‑on, trade‑based work in the low‑risk category highlights a crucial limitation of today’s AI: while algorithms excel at processing large datasets and executing predefined instructions, they falter when faced with ambiguous, spatially varied, or physically demanding tasks. Jobs that require workers to manipulate tools, assess material quality, respond to unexpected hazards, or customize solutions to unique site conditions continue to rely heavily on human perception, motor skills, and experiential knowledge. This reality offers a degree of insulation for those professions, at least until breakthroughs in embodied AI and advanced robotics close the gap.
Recognizing the gendered dimensions of AI’s impact is essential for crafting effective policy responses. Without an explicit gender lens, governments and industry risk overlooking the disproportionate burden that automation may place on women workers, potentially exacerbating existing wage gaps and limiting career advancement opportunities. Fortunately, Australia already mandates that all new policy proposals undergo a gender impact assessment. Extending this requirement to the newly created Office of AI would ensure that any regulatory framework, funding initiative, or workforce‑development program accounts for how automation affects different gender groups.
Applying a gender‑aware approach to AI policy can transform a looming challenge into an opportunity for inclusive growth. Policymakers can design targeted upskilling pathways that help clerical and administrative workers translate their strong organisational, communication, and data‑management abilities into emerging AI‑augmented roles such as AI‑assisted process analysts, digital compliance officers, or data‑quality specialists. Simultaneously, investments in lifelong learning can support transitions into hybrid positions that combine human judgement with AI tools—for example, supervising automated payroll systems while focusing on exception handling and employee wellbeing.
Beyond individual reskilling, a gender‑informed strategy calls for broader labour‑market reforms that promote equity and resilience. Employers should audit their internal talent pipelines to identify roles at high risk of automation and create internal mobility programs that prioritize affected workers for retraining. Educational institutions can partner with industry to develop micro‑credential programs focused on AI literacy, prompt engineering, and ethical AI use, ensuring that women have equitable access to these emerging skill sets. By proactively matching transferable competencies with new job categories—both those created by AI and those that remain inherently human—Australia can build a labour market that adapts swiftly while advancing gender equity.
In conclusion, the evidence clearly shows that AI’s disruptive potential is not distributed evenly across the workforce; it converges on sectors where women are heavily represented, particularly clerical and administrative support. Recognizing this pattern enables stakeholders to move beyond generic forecasts and implement precise, actionable measures that protect livelihoods, foster skill development, and promote an inclusive economic future. Workers in exposed roles should seek out AI‑literacy training and consider how their core strengths—organisation, attention to detail, and interpersonal coordination—can be re‑applied in tech‑enhanced environments. Employers must invest in reskilling and internal mobility, while policymakers need to enforce gender‑impact assessments for all AI‑related initiatives. Together, these steps will help transform a potential disruption into a catalyst for a more adaptable, fair, and prosperous labour market.