Across high-adoption Australian states, how does tying a share of university and major-hospital AI funding to measured gap‑closing in per-capita work and coursework AI use at co-located TAFEs, RTOs, councils, and community health services change (a) adoption concentration between high- and low-status institutions and (b) the local use-case mix, compared with hub-focused funding that has no explicit gap-closing requirement?

Answer

Tying part of hub funding to local gap-closing probably reduces adoption concentration and shifts the use-case mix in lower-status institutions toward more work and coursework, but effects depend on how tightly metrics and supports are designed.

(a) Adoption concentration

• With pure hub-focused funding (no gap rule):
  • Universities/major hospitals keep pulling ahead on per-capita use.
  • Local TAFEs, RTOs, councils, community health lag; gaps often widen.
• With gap-linked funding:
  • If a visible share of hub funding is conditional on narrowing per-capita gaps with named co-located LSIs, hubs gain incentives to:
    • Share licences/platforms.
    • Co-develop templates and training.
    • Place staff time into LSI support.
  • Likely pattern:
    • Faster growth in LSIs than under status quo.
    • Smaller per-capita gaps within each locality.
    • State-level adoption concentration still present, but less skewed toward a few universities/hospitals.
  • Risk:
    • If measures are weak or easy to game (e.g., counting one-off workshops), hubs may meet targets without durable LSI use; concentration barely changes.

(b) Local use-case mix

• Baseline (no gap rule):
  • Hubs: growing mix of research, clinical/teaching, and admin.
  • LSIs: mostly light admin and some coursework; limited frontline work.
• With gap-linked funding plus simple conditions:
  • TAFEs/RTOs:
    • More course-embedded use (assessments, practice tasks) as hubs co-develop units and guardrails.
    • Some spillover to student work-use where tools are allowed for placements/apprenticeships.
  • Councils/community health:
    • Slight shift from generic admin toward standardised frontline templates (intake, follow-up letters, information sheets) if these are counted in metrics.
  • Overall:
    • Work and coursework shares in LSIs rise relative to personal/experimental use.
    • Mix in hubs changes little, except more activity in outreach/teaching for LSIs.
  • Risk:
    • If metrics count any “AI activity”, LSIs may chase low-value admin use; mix shifts less than hoped.

Design implications

• Make funding contingent on:
  • Per-capita work and coursework use in each named LSI growing at least as fast as in the hub.
  • A minimum share of measured tasks in LSIs tagged as work or coursework, not just admin.
• Provide enablers:
  • Shared platforms and templates.
  • Pooled assurance so LSIs can safely adopt frontline/course uses.
• Keep hub incentives local:
  • Tie a portion of each university/hospital’s AI budget to outcomes for specified neighbouring LSIs, not generic “regional outreach”.