How Australia Uses Claude: Findings from the Anthropic Economic Index

Anthropic has announced that Australia is among the leading global adopters of Claude, with per capita usage more than four times higher than predicted by its working-age population. This high adoption is coupled with a distinct usage profile characterized by diverse, non-technical tasks and a collaborative approach to AI interaction.

High Per Capita Adoption and Geographic Concentration

Australia ranks seventh globally in per capita Claude adoption, with an Anthropic AI Usage Index (AUI) of 4.1. While the country accounts for 1.6% of global Claude.ai traffic, usage is heavily concentrated in specific regions:

• New South Wales: Accounts for 37.2% of conversations and has an AUI of 1.20.
• Victoria: Accounts for 30.8% of conversations and has an AUI of 1.19.

Every other state and territory has an AUI below 1, with the lowest adoption found in the Northern Territory (0.12) and Tasmania (0.32).

Unlike global trends where income strongly predicts AI adoption, state-level adoption in Australia is driven by workforce composition. High-income, mining-heavy regions like Western Australia and the Northern Territory show low usage, while New South Wales and Victoria—which have slightly below-average income—show high adoption due to a higher concentration of workers in finance, professional services, and tech sectors.

Usage Profiles and Anglosphere Alignment

Australia's use of Claude closely mirrors other high-income Anglosphere economies in terms of use-case distribution and interaction style:

• Use-Case Mix: 46% of conversations are for work, 47% for personal use, and 7% for coursework. This low share of coursework is typical of high-adoption, high-income economies.
• AI Autonomy: Australia has a relatively low AI autonomy score of 3.38 on a 1–5 scale, indicating that users prefer collaborative interaction over full delegation of decision-making to the model.
• Task Complexity: Australian prompts are sophisticated, requiring an estimated 11.9 years of schooling to understand. However, the tasks themselves are relatively less time-intensive, taking an estimated 2.7 hours to complete without AI, compared to the global average of 3.3 hours.

Diversity of Tasks and Technical Divergence

Australia exhibits a more diverse range of Claude usage than the global average and its Anglosphere peers. While the top 100 tasks in the US, UK, and Canada account for roughly 48–50% of usage, Australia's top 100 tasks account for only 47.3%.

This diversity is primarily driven by a lower reliance on coding-related tasks compared to the global baseline:

• Underrepresented Tasks: Computer and Mathematical tasks are 8.0 percentage points below the global average. General coding assistance (13.5% of Australian use) is significantly lower than the global average (16.8%).
• Overrepresented Tasks: The gap left by coding is filled by a broader array of non-technical professional and personal tasks, including Management (+2.3pp), Office and Administrative Support (+1.3pp), and personal life management (+1.9pp).

Australia's divergence from its Anglosphere peers is most notable in its lean toward management and administrative support over educational instruction, which is 2.7 percentage points below global parity.