Millions of Australians go to work each day unaware that their role is quietly being reshaped by automation. Using occupation level data, we can now see which jobs face the highest risk, who is most exposed, and whether employment figures are already starting to shift.
Some occupations are far more exposed than others. Clerical, administrative, and routine manual roles consistently sit at the top of automation risk rankings. Hover over each bar to explore the data.
Figure 1. Automation risk by occupation group. Source: Frey & Osborne (2017).
When automation risk is plotted against average weekly wages and total employment size, a troubling pattern emerges: occupations with the highest automation risk tend to have the lowest wages and employ the most people. These are not niche roles, they are the backbone of the working class. Bubble size reflects the number of workers in each occupation, using the dataset as an occupation level risk benchmark.
Figure 2. Automation risk vs average weekly wage vs employment size. Source: Dataset adapted from Frey & Osborne occupation risk estimates and US labour market variables. Frey & Osborne (2017).
Comparing employment trends in high-risk versus low-risk occupation groups over the past decade reveals a clear divergence. High-risk occupations have stagnated or declined while low-risk roles have grown steadily. Toggle the groups on and off using the legend.
Figure 3. Employment trends by automation risk group, 2014–2024. Source: ABS (2024a).
While some industries are shedding routine roles, others are rapidly building demand for AI skills. The chart below shows how AI skill requirements in Australian job postings have grown across sectors since 2018. Hover over each bar to compare industries and years.
Figure 4. Share of Australian job postings requiring AI skills by industry and year. Source: PwC (2025).
Frey and Osborne’s original data reveals a clear relationship: the more education an occupation typically requires, the lower its automation risk. But some mid level education occupations still face significant exposure and they linked to large employment groups in the dataset. Hover over each point to compare education groups.
Figure 5. Education level vs average automation risk by education group. Source: Dataset adapted from Frey & Osborne (2017).
The data tells a consistent story: automation risk falls hardest on workers who are already the most economically vulnerable, those in lower paid, lower education roles that employ large numbers of Australians. At the same time, demand for AI skills is growing rapidly, but concentrated in industries that already pay well.
The question is no longer whether automation will reshape Australian work. It is whether the workers most at risk will have the policy support, retraining pathways, and economic buffer to adapt before the wave reaches them.
Anthropic’s Claude LLM was used to teach certain RStudio skills that I was not confident in and supported in debugging. All research, code, and writing submitted was that of my own.
Australian Bureau of Statistics. (2024a). Labour force, Australia, detailed, November 2024. ABS. https://www.abs.gov.au/statistics/labour/employment-and-unemployment/labour-force-australia-detailed/latest-release
Frey, C. B., & Osborne, M. A. (2017). The future of employment: How susceptible are jobs to computerisation? Technological Forecasting and Social Change, 114, 254–280. https://doi.org/10.1016/j.techfore.2016.08.019
PricewaterhouseCoopers Australia. (2025). The fearless future: How AI is impacting Australia’s jobs and workers. PwC. https://www.pwc.com.au/services/artificial-intelligence/ai-jobs-barometer-report-2025.pdf
Plotly. (2026). job-automation-probability.csv [Data set]. GitHub. https://github.com/plotly/datasets/blob/master/job-automation-probability.csv