Blink and you may have missed it amid the torrent of tech news, but Australia’s chief science body, the CSIRO, earlier this month released an interesting bit of research into AI.
Analysing job advertisements from over 4,000 companies, it found that firms using AI were not broadly killing jobs in response to the rise of the technology, but in fact were searching for more employees in order to properly harness it.
It’s the counter narrative to what we’ve been hearing more broadly. When the CSIRO speaks on a matter, it’s usually with fact and authority. It’s debunked many myths over the years, from diet fads through to archaeological discoveries.
Yet, in this instance, when the subject is so incredibly topical, the finding fell on deaf ears.
The main line of attack I’ve seen on social media is that this research focused on job advertisements over actual hires. Though, economically, job advertisements have historically been used to indicate the strength and forecast of the economy. It’s significant data.
Yet, it seems the narrative around AI destroying more jobs than it is creating is so entrenched that not even science can stop it.
All of this has me thinking more broadly about how AI will impact the future of the jobs market. As the founder of an insurance technology business leveraging this technology, I can see it has the potential to create new roles and change the way we work.
Yet, there’s a strong undercurrent to the contrary.
While visiting Australia recently, Microsoft’s Satya Nadella warned yet again of a job upheaval triggered by AI. More broadly, we’ve seen companies like Block and Atlassian shed roles en masse, crediting the technology as the cause.
The paradox of automation
The “AI as job killer” story is sticky for a reason. Automation anxiety has a long history, and every new wave of technology brings the same fear: a machine steps in, a human steps out.
What people consistently get wrong is conflating the automation of specific tasks with the elimination of jobs wholesale.
There’s an economic concept worth understanding here: Jevons Paradox. When a process becomes faster and cheaper, demand for it tends to explode rather than contract. Think about what spreadsheet software did to accounting.
It didn’t hollow out the profession—it made financial modelling so accessible that demand for deep financial analysis boomed, and the industry ended up needing more accountants, not fewer.
The CSIRO data suggests the same dynamic is playing out right now. Firms that had adopted AI posted 36% more non-AI job advertisements than comparable firms that hadn’t. The real divide isn’t humans versus machines. It’s companies embracing the technology versus those sitting on their hands.
At upcover, I see this daily. We don’t use AI to cut headcount. We use it to remove repetitive work so our people spend their time on problems that actually require a human brain.
Take what’s happening in legal. AI can draft a contract in seconds. But because generating that document is now nearly instantaneous, the review and approval stage becomes the bottleneck, and you need more legal professionals to manage the volume coming through the door, not fewer. The output changes; the need for human judgment doesn’t go anywhere.
New roles, not fewer roles
What does shift are the roles themselves. We’re already seeing entirely new jobs emerge—AI product managers, agent operators, people whose function is directing, auditing and quality-controlling what the machines produce.
These didn’t exist five years ago. The CSIRO research picked up on this too: AI-related skills are appearing in job ads for sales representatives, security officers, architects. The line between an “AI job” and a regular job is already blurring.
But here’s the problem I don’t think gets enough attention. The traditional career ladder runs on an apprenticeship model. Junior lawyers proofread standard contracts. Junior developers write basic code. They learn by doing the heavy lifting. If AI absorbs all of those entry-level tasks, how do you cultivate the next generation of senior people?
Companies that treat AI purely as a mechanism to thin out junior ranks are trading a short-term efficiency gain for a long-term talent shortage.
The firms that get this right won’t eliminate those roles—they’ll redesign them from the ground up. Junior staff should be directing AI tools, auditing AI-generated output, and developing the judgment that separates good work from bad. That requires a different kind of mentorship, one that prioritises quality control and critical thinking over basic execution.
For anyone entering the workforce right now: ignore the doomsday headlines.
The CSIRO data backs what those of us building companies actually see in practice. The people being disadvantaged by AI are not those working alongside it, but those stuck in organisations that aren’t using it at all.
The technology is not the threat. Complacency is.
- Anish Sinha is co-founder of upcover, a specialist AI-enabled insurance platform.