If we reclassify Australian AI adoption not by geography first but by access channel—individual licences, organisational licences, and mediated/shared access points (libraries, unions, community hubs, industry portals)—how different do regional vs metropolitan gaps in adoption concentration and use-case mix look, and which current state or federal deployment targets become clearly misaligned with where person-level AI use is actually growing?

Answer

Reclassifying by access channel likely shrinks some metro–regional gaps and reveals that growth is strongest in mediated/shared channels that current targets mostly ignore.

Channel effects on metro–regional gaps

• Individual licences

  • Metro: higher density, skewed to professionals and university students; more work and coursework use.
  • Regional: fewer paid seats; heavier reliance on free tools; mix leans personal/admin.
  • Net: geography still matters; gaps look similar to standard maps.

• Organisational licences

  • Metro: concentrated in universities, major hospitals, central agencies, large firms.
  • Regional: some use in health, councils, schools, mines, agri‑firms, but many small orgs excluded.
  • Net: adoption concentration looks even more hub‑heavy when viewed this way; state targets that count org licences overstate diffusion.

• Mediated/shared access (libraries, TAFEs, unions, community hubs, industry portals)

  • Metro: used but overshadowed by individual/org seats.
  • Regional/outer suburban: relatively more important; early growth in person‑level use for gig work, micro‑business, study support.
  • Net: when weighted by people reached, mediated channels narrow regional gaps in light use, especially for work and coursework boosters at low income.

Implications for adoption concentration and use‑case mix

• Concentration
  • By licences: still metro‑heavy, dominated by a small number of urban orgs and micro‑clusters.
  • By people actually using AI weekly: the map flattens somewhat once mediated channels and free tools in regional settings are included.
• Use‑case mix
  • Individual + org licences: metro users do more complex work/course tasks; regional users with seats skew to generic admin and compliance.
  • Mediated access: both metro and regional users lean toward income‑related micro‑tasks, coursework support, and basic admin; personal/entertainment uses are present but not dominant among repeat users.

Misaligned deployment targets

• Licence-count targets

  • Overweight organisational seats in metros; undercount mediated/shared access that drives first steps in regional and low‑status settings.
  • Misalignment: programs can “hit” a target while person-level use grows mainly in already‑dense metro hubs.

• Hub/pilot concentration

  • State and Commonwealth pilots in universities, capital‑city hospitals, and central agencies treat those orgs as primary diffusion engines.
  • Reclassified view shows many regional users first touch AI via libraries, TAFEs, industry portals, unions, and platforms, not via local copies of metro hubs.

• Equity and regional targets

  • Current equity framing often uses geography and institutional type (e.g., “regional universities”, “regional schools”).
  • Channel lens suggests targets should track: (a) per-capita mediated-access use, and (b) growth in shared channels serving time-poor, precarious workers and learners.

Most actionable shifts

• Add explicit per-capita targets for mediated/shared access in regional and outer-suburban areas, not just counts of org licences.
• Tie some hub funding to demonstrated growth in mediated or shared-use channels in their catchments (e.g., TAFEs, libraries, small-firm portals).
• Reweight public-sector deployment KPIs from “agencies onboarded” to “staff using via org or shared channels per 100 FTE, by region”.