The CSIRO Data That Undercuts the AI Redundancy Story

The headlines have settled into a comfortable narrative: adopt AI, cut jobs, book the savings. It is tidy, it is scary, and it makes for good copy. It also does not match what Australian employers have actually been doing.

CSIRO has put some local numbers on the table, and they point the other way.

What the research actually found

CSIRO analysed a national dataset of online job ads from more than 4,000 Australian firms, identifying which ones showed signs of adopting AI before 2020 and then tracking both groups over the following three years. The study, published in the Australian Journal of Labour Economics, analysed a national dataset of online job advertisements from more than 4,000 Australian firms, identifying AI adopting and non-adopting firms based on signals in job postings prior to 2020 and then comparing the two groups over the next three years.

The headline result is the one worth pinning to the wall. After accounting for factors such as firm size, industry and location, AI-adopting firms posted 36 per cent more non-AI job ads over time than non-adopting firms. In other words, firms leaning into AI were hiring more people, not less.

The roles people usually worry about held up too. The AI-exposed roles did not see a decline in demand in firms that adopted AI. Instead, the slight dip for AI-exposed workers happened in firms that were not adopting the technology. As study lead Dr Claire Mason put it, that suggests AI-exposed workers may be disadvantaged if they are in firms that are not using AI, while their peers in AI-adopting firms are potentially more competitive because they are able to use these tools to augment their work.

A fair caveat: this is a correlation, and the data covers 2020 to 2023. CSIRO's research looked into trends between 2020 and 2023, before generative AI became widely used in organisations, but the agency noted that more recent Australian labour market data are demonstrating similar patterns. The researchers are careful about it. The study does not claim AI is risk-free or that disruption won't occur, but it offers grounded, local evidence that AI adoption has been associated with stronger demand for workers and richer skill profiles, rather than mass displacement.

The skills story is the bigger story

Hiring more people is only half of it. The nature of the roles shifted. Across the dataset, job ads began listing more skills over time, with the increase being strongest in AI-adopting firms and in AI-exposed roles. The read on that is worth sitting with. In many ways the data counters commonly held fears about deskilling; what we are seeing is workers being asked to bring more skills to the table, including the ability to work effectively with AI.

And this is not just a tech-team phenomenon. AI-related skills were not confined to technical or IT-heavy roles, and were starting to appear in jobs you might not expect, from sales representatives to security officers and architects. The upshot for how you think about your org chart: the distinction between 'AI jobs' and 'non-AI jobs' is starting to blur, with AI becoming a core part of many roles rather than a separate category.

This sits alongside the national picture. The Generative AI Capacity Study released by Jobs and Skills Australia in August 2025 concluded that 79 per cent of Australian workers face low or very low risk of automation from generative AI, with the technology primarily enhancing rather than eliminating roles. Two independent Australian sources, same direction of travel.

How to position AI internally

None of this means the redundancy talk vanishes on its own. It means you have credible local evidence to counter it, and a clearer job to do with your own people. A few practical moves.

Be direct about the goal. If your AI investment is really about growth, capacity and doing better work, say so plainly and back it with where the freed-up time is going. If it is genuinely about cost, do not dress it up as something else, because your staff will read the gap between the words and the actions.

Frame AI as a skill your people build, not a thing that happens to them. The CSIRO finding is that the people and firms using AI well are the ones pulling ahead. Give staff the training, the time and the permission to get good at it.

Watch for the quiet risk in the data. The soft spot was not AI-adopting firms cutting roles, it was non-adopting firms where AI-exposed workers slipped behind. Standing still is its own kind of decision, and it is not the safe one it feels like.

Pick roles where AI adds capacity to people you already have, rather than starting from a headcount target. That order tends to produce both better work and a workforce that trusts the change.

The right next steps for you depend on where you are right now. Our AI Maturity Assessment is an easy place to start. After that, let's have a chat about your next steps.