The prime minister’s recent AI policy announcement has shoved artificial intelligence into the centre of government power, with new Australian standards, an Office of AI inside the prime minister’s own department, and expectations that major AI companies must “earn” their social licence to operate locally.
That’s the official story. The real machinery sits lower down, in the hands of the people expected to design, procure, deploy, audit and maintain the systems while Canberra writes the rules and the companies chase profit.
Who Holds the Levers
Dr Prins Ralston, chief executive of the Australian Computer Society, said the whole framework rises or falls on workforce capability. “Every obligation the prime minister now wants to impose – from copyright protection and social licence, to grid resilience and national security – will ultimately live or die in the hands of the workforce that designs, procures, deploys, audits and maintains AI systems,” he said.
He added that treating workforce capability, professional recognition and sovereign credentialing as a downstream execution detail would be “a fundamental error.” In his words, “In the AI era, they are critical national infrastructure.” That’s a neat phrase for a system where the state sets the terms, the firms extract the value, and workers are left to make the whole thing function.
The article said the Office of AI inside the prime minister’s department signals that AI has moved from a niche innovation issue to a whole-of-economy agenda touching jobs, energy, defence, copyright and productivity. In other words, the apparatus is expanding. So is the reach.
Who Pays for the Upgrade
The ACS Australia’s Digital Pulse 2026 report, produced by Deloitte Access Economics, said digital skills contributed an estimated $476 billion to the economy in 2025. The tech sector added $142 billion, and exports exceeded $12 billion. Those numbers sound grand from the top floor. Down below, the report said technology workers are at the forefront of AI disruption, with 70 per cent of their work time now affected by AI through automation, augmentation or entirely new responsibilities.
The report also said 40 per cent of new tech workers over the past five years entered through non-traditional routes, including workplace training, industry-recognised credentials and self-starter pathways with no formal technology qualifications. That diversity, the report said, helps address long-standing shortages and improves gender diversity, with women outnumbering men in some alternative pathways.
So the system leans on workers it didn’t fully train, then calls that flexibility.
Ralston said professionalisation means more than technical know-how. “Professionalisation requires someone who looks at technology from a business perspective and a security perspective, not just from an IT perspective,” he said. “The result is someone who understands the big picture and is focused on delivering business outcomes, not just deploying technology.”
Credentials, Wages, and Control
The report found that accredited training mapped against common skills frameworks such as the Skills for the Information Age, or SFIA, is associated with a wage premium of around 7.3 per cent, or $8600 per year for the average technology worker. In AI-exposed areas such as cybersecurity, data and machine learning, it said consistent ways to assess capability, ongoing professional development and frameworks such as SFIA are even more important.
That’s the language of managed access. Recognition, credentialing and professional standards become the gates through which workers must pass to be considered legitimate enough to keep the system running.
Ralston put the point bluntly: “Policy execution, in this context, means people.” He said, “If the AI framework does not build a coherent system for recognising, credentialing and mobilising domestic capability, Canberra will find itself with beautifully drafted rules and too few trusted professionals to implement them.”
He also said the Office of AI must build workforce capability, professional recognition and sovereign credentialing into the framework from the start if it wants the economic and productivity gains it has been tasked with delivering.
Then came the warning. “Otherwise, Australia risks becoming exactly what the prime minister warns against: a ‘data warehouse’ for AI products made elsewhere, serviced by a thin layer of domestic implementation capacity,” Ralston said.
That’s the whole arrangement in one line: centralised policy, corporate systems, and a workforce expected to absorb the risk, the labour and the fallout. Capital follows clarity, he said. In the age of AI, clarity about human capability will be as important as clarity about standards and regulation. The people at the bottom, once again, are the ones expected to make the machine work.