Key Points:
- In Australia, 0.08% of job postings mention GenAI-related keywords in their job descriptions. That number has increased rapidly but trails our global peers including the United States, the United Kingdom and Canada.
- Around one in five Australian jobs are highly exposed to potential change from GenAI tools. These tools can execute at least 80% of the skills required to perform these high-exposure roles at a ‘good’ or ‘excellent’ level.
- The highest potential exposure to GenAI can be found across a wide variety of knowledge work, particularly in the tech sector. The smallest potential exposure to GenAI can be found in more physically intensive roles, including driving, nursing and construction.
Every job across the Australian labour market is potentially exposed to some degree of change from generative artificial intelligence (GenAI), ranging from relatively low exposure in driving or nursing roles to high exposure across a range of tech occupations. Whether exposure to GenAI proves beneficial or not to the overall labour market remains to be seen, but right now it seems GenAI is unlikely to fully replace many jobs. Instead, depending on the professional skills GenAI is comparatively ‘good’ or ‘bad’ at, it may only augment some functions of existing jobs, with some jobs transforming more than others.
To quantify the exposure of Australia’s job market to potential GenAI-driven change, the Indeed Hiring Lab examined millions of job postings and identified thousands of professional skills. We then asked ChatGPT, a leading GenAI tool, to evaluate its own ability to perform a range of job-relevant skills on a scale from ‘poor’ to ‘fair,’ ‘good,’ or ‘excellent.’ To gauge how many roles currently require GenAI skills — either to use the tools or help develop them — we also analyzed the share of Australian job postings that mention any of several GenAI-related keywords.
GenAI and the Australian job market
Just a year or so ago, few Australians had even heard of GenAI, let alone tried to hire people with expertise in its usage or development. But while job postings relating to GenAI remain a relatively small part of the market — just 0.08% of Australian job postings mentioned GenAI in their job descriptions as of the end of March 2024 — they are increasing rapidly. The share is up by around 50% over the past three months.
Nevertheless, Australian workplaces appear to be embracing GenAI more slowly than our global peers, even when accounting for the existing size of each country’s AI sector. The share of job postings mentioning GenAI is significantly higher in Singapore (0.53% of postings) and other large, wealthy economies including the United States (0.12%), the United Kingdom (0.11%) and Canada (0.15%).
How exposed is Australia’s labour market to ChatGPT?
We found that around one in five (21%) Australian job postings on Indeed face a ‘high exposure’ to GenAI. That is, GenAI tools such as ChatGPT can perform at least 80% of the skills required in those jobs at a ‘good’ or ‘excellent’ level. A further 56% of job postings had a ‘moderate exposure’ (GenAI could perform between 50% and 80% of the skills at a high level). Australia’s results are similar to those observed in the United States, particularly for high exposure, with moderate exposure somewhat higher in Australia.
This analysis of Australia’s exposure to GenAI reflects the scope and capacity of GenAI tools today. Future exposure to GenAI will depend on a number of factors, including the evolution of GenAI itself and the changing composition of Australian job creation. Presumably, the competency of GenAI tools will improve over time, with some skills becoming more easily performed.
The occupations with the lowest exposure tend to be quite physically intensive roles, including driving, nursing and construction. GenAI can reasonably perform at least some of the skills required by each of these jobs, but typically few of the critical skills. For example, GenAI can perform just 29% of the skills required in a driving occupation, and 39% of skills required for nurses, at a ‘good’ or ‘excellent’ level. In these occupations, GenAI might help to manage some routine or administrative tasks but is largely unhelpful in performing the core tasks required.
By comparison, the occupations with the highest exposure include a range of white-collar roles. Tech seems particularly exposed, with software development, IT operations and information design all featuring prominently on the list of roles that GenAI can (theoretically, at least) perform at a high level. For example, GenAI can perform 96% of the skills required in a typical software development role at a ‘good’ or ‘excellent’ level. The back-office of workplaces everywhere could also be transformed by GenAI, with legal, accounting, human resources and marketing roles all potentially highly exposed to these new tools.
It is important to note that high exposure to GenAI isn’t necessarily a bad thing. Workers in potentially highly exposed jobs may actually receive a greater benefit from these tools, which could boost productivity by automating some mundane tasks and free up workers to perform more important or critical tasks. Research from the Harvard Business School found that consultants at the Boston Consulting Group who used ChatGPT completed tasks 25% faster, and their work quality was assessed to be 40% higher, than the control group of colleagues who were not AI-assisted. Since productivity growth tends to drive wage growth, this could create a divide in wage growth between high- and low-exposure occupations, with the former enjoying stronger wage gains on the back of GenAI-enhanced productivity.
As GenAI usage becomes more prevalent, tracking growth in high- versus low-exposure occupations over time will be important to understanding whether GenAI is having a positive or negative impact on particular occupations.
Conclusion
Unlike earlier technological advances in robotics and computing that primarily replaced or reduced the demand for manual labour, it is knowledge workers who will be most affected by the emergence of GenAI. Whether this impact proves to be positive or negative remains to be seen, but GenAI does appear likely to have a transformative effect on some occupations, particularly in areas such as tech and throughout a range of back-office roles.
Right now the Australian labour market isn’t highly exposed to GenAI, but that is very likely to change over time as GenAI tools evolve and the composition of the Australian job market adapts to the usage of these tools. Our analysis provides a useful snapshot of how the situation looks right now, but we should be mindful of how quickly that can change.
Methodology
The analysis involved extracting job postings directly related to Generative AI, using specific keywords indicating its presence, such as ‘Generative AI,’ ‘Large Language Models,’ and ‘ChatGPT.’ For our methodology on country exposure levels and the most- and least-exposed occupations, please see our post here.
