Section
Australian AI research, adoption and labour-market evidence, taken from the reports and datasets themselves, every figure linked to its source with the date we read it, caveats included.
Research
Answers to the Joint Select Committee on AI: the Productivity Commission’s $116 billion is not a cost-benefit analysis, nobody has modelled AI job losses or offshore payments, and the workplace department holds no data on whether employers consult workers before bringing AI in.
Data
What the 139,000 in the Jobs and Skills Report 2026 actually measures, the box that puts AI outside the baseline projections, and what the report’s own evidence on AI and work says.
Research
The AI Safety Institute’s CSIRO report on scalable oversight: what its AI-on-AI experiments tested, the training condition its clear steering result depends on, and what it means for anyone relying on a human to check an AI.
Data
eSafety's survey of 1,950 children, taken before the Age-Restricted Material Codes for AI services began: 78 per cent had used an AI assistant and 8 per cent a companion, assistant users sought personal advice more often than companion users, and one in five who had used either had a potentially harmful interaction.
Data
The OAIC's triennial attitudes survey puts AI companies equal-last on trust with data brokers. Ninety-six per cent want conditions before an AI decides anything about them, demand for accuracy validation and bias testing rose fastest of all, and only 25 per cent accept AI for eligibility decisions.
Data
DEWR's monitoring framework finds exposed occupations growing 5.6 per cent against 9.5 for the least exposed, says itself the result is not definitive, and shows the most exposed workers are majority women with degrees.
Data
NAIC and CSIRO's ecosystem census: 25 clusters led by Melbourne CBD, patents near-quadrupled, research share of global output falling, and a commercialisation gap the report calls out itself.
