Nearly half of Australian adults have now used generative artificial intelligence at least once, according to new research from the Australian National University conducted in partnership with Google, giving policymakers their clearest read yet on how deeply the technology has moved into everyday work and study.
The report, titled AI Adoption in Australia, surveyed more than 3,500 adults and interviewed senior leaders at organisations adopting the technology. It found that 48.6 per cent of respondents have used a generative AI tool such as ChatGPT, Google Gemini, Dall-E or Claude. The figure lands generative AI somewhere between a novelty and a fixture, roughly six years after the first large language models reached the public.
Study lead Dr Jessica Herrington, from the ANU School of Cybernetics, said the technology had “moved rapidly from a niche technological development to a widely recognised and increasingly used tool,” according to the ANU’s news release. Professor Nicholas Biddle, head of the ANU School of Politics and International Relations, was a co-author.
Who is using it, and who is not
The headline number hides a sharp split. The ANU researchers found adoption is much higher among younger Australians, people with tertiary qualifications, and those living in capital cities. That pattern points to what the report calls “emerging digital capability divides” rather than uniform take-up.
Earlier ANU analysis, reported by the Australian Computer Society’s Information Age, put hard numbers on that gap. Among 18-to-34-year-olds, recent generative AI use ran at 69.1 per cent; among 65-to-74-year-olds it fell to 15.5 per cent. Students recorded the highest usage of any group at 78.9 per cent, while people who had not finished high school sat near 20 per cent. Professionals used the tools at more than double the rate of machinery operators.
The composition of use is narrow, too. Information Age reported that text generation dominated at 82.6 per cent of users, well ahead of image generation and code generation. In other words, most Australians who have tried generative AI are using it to write, summarise and draft, not to build.
Training, not curiosity, is the ceiling
Where adoption stalls, the ANU report is specific about why. The leading barriers were insufficient training and skills, concerns about how personal data is handled, and a fear of over-reliance that could erode critical thinking. Curiosity is not the constraint; capability is.
That matters because the users who have crossed the threshold are largely positive. Most reported time savings as the primary benefit, freeing capacity for more complex or creative work. Students and workers said the tools improved the quality of their study and their job performance, per the TechXplore account of the research and the ANU’s own summary.
Nearly 75 per cent of respondents said they had at least a basic or moderate understanding of the technology. The gap between that self-assessed literacy and the training barrier suggests many Australians feel they grasp what generative AI does but lack structured support to use it well, safely, or at work.
A public that wants guardrails
Australians are not sanguine about where this is heading. The ANU found 66.2 per cent believe the government should regulate generative AI, and 48.1 per cent support an independent authority to oversee it. Concern spikes around civic life: 79.4 per cent said they were concerned or very concerned about the technology’s role in politics, with misinformation and the handling of political data among the worries.
Dr Herrington called for targeted training, transparent governance, consistent labelling of AI-generated content, and coordinated action across government, industry and civil society. Senior leaders interviewed for the report described the technology as transformational but said they wanted clear standards and shared responsibility frameworks from providers.
Why it matters for Australia
This is the most current and robust national picture of generative AI use in Australia, and it arrives as the country moves from debate to delivery. The finding that a lack of training, not disinterest, is the core barrier is a direct brief for the programmes now rolling out.
The federal government has begun a national AI literacy programme for schools this year, aiming to reach one million students within three years with curriculum-aligned modules, and is offering one million fully subsidised AI microskill scholarships through the National AI Centre and TAFE NSW, as detailed by the Minister for Industry and Innovation. The ANU data quantifies the demand those schemes are meant to meet, and shows which cohorts, older Australians, people without tertiary qualifications, and workers in trades and operations, risk being left on the wrong side of the divide.
For educators and training providers, including the TAFE and registered training sector, the study reframes AI literacy as workforce infrastructure rather than an optional extra. For policymakers, the two-thirds who want regulation and the near-four-fifths worried about AI in politics signal a public that will accept, and may expect, firmer rules.
The unresolved question is whether Australia’s literacy push can close a gap that is still widening. Adoption is climbing fastest among those already advantaged by age, education and location. Unless training reaches the groups the ANU flags as lagging, the next survey may show a larger total using generative AI and a deeper divide over who benefits from it.
Sources: Australian National University, TechXplore, ACS Information Age, Minister for Industry and Innovation.
