While generative AI could result in greater automation of some roles, other knowledge-based roles are more likely to be augmented than replaced, said SEEK chief economist Blair Chapman.
"AI may be increasing demand for junior workers in some white-collar roles, particularly where it helps workers do tasks more effectively rather than replacing them outright," Dr Chapman said.
"In other roles, AI appears to reduce demand by doing some tasks itself. This suggests the labour market effects of AI are still unfolding."
Dr Chapman used SEEK job ad data and AI exposure measures from Jobs and Skills Australia to determine which occupations would be augmented by AI and which would have their roles automated by it.
Because the rise of mechanisation and computerisation has already reduced the share of employment in routine jobs, generative AI is most likely to have an early observable impact in non-routine cognitive roles.
Jobs that rely on language, analysis, communication or judgement are more likely to be impacted.
SEEK job ads mentioning AI-related skills grew strongest in IT, marketing and consulting between 2019 and 2026. Healthcare, construction and hospitality were largely unaffected.
Swiss banking giant UBS recently included questions about AI fluency in graduate recruitment interviews, becoming one of the first major financial institutions to stipulate AI literacy skills for entry-level roles, according to a report in the Financial Times.
Of the top 15 non-routine cognitive occupations in Australia in 2025, only software and applications programmers had an automation exposure above 60 per cent.
It recorded a 0.3 percentage-point decline in the early-career share of job ads.
On the other end of the spectrum, electronics engineers had an augmentation exposure score of 76 per cent and an automation exposure of 34 per cent.
It recorded a three percentage-point increase in the early-career share of job ads.
Other jobs with relatively high augmentation exposure and relatively low automation exposure included civil engineers, advertising professionals and construction managers.
Most non-routine cognitive occupations had higher augmentation than automation scores, meaning generative AI was more likely to assist with tasks rather than fully automate them, Dr Chapman said.
"This suggests that, for now, GenAI is more closely associated with increased demand for early-career workers than with reduced demand," he said.
But it does not necessarily mean demand for workers in general will rise.
"If AI helps people work much more efficiently, employers may still need fewer workers overall. But it could also change the mix of workers employers need," Dr Chapman said.
"For example, if GenAI makes less experienced workers more productive, employers may be able to hire more junior staff to do work that previously required more experience."
