The gap between AI adoption and AI capability
Australian organisations are adopting AI at accelerating pace — the 2025 Microsoft Work Trend Index found 37% of Australias working-age population using generative AI tools by the end of 2025. But adoption is running well ahead of capability. Tools are being purchased; capability is not being built. The predictable consequences: AI outputs submitted without adequate review; confidential data entered into consumer AI tools; AI-generated errors nobody caught because nobody knew what to check for.
What the national frameworks say
The National AI Plan (December 2025) committed one million fully subsidised microskill scholarships through TAFE NSW and the National AI Centre. The APS AI Plan (November 2025) mandated foundational AI literacy for all APS staff, established Chief AI Officers in every Commonwealth agency, and required agencies to develop strategic positions on AI adoption. Microsoft announced in May 2026 a commitment to skill three million Australians by 2028. These set the external bar for what organisational investment is expected to look like.
The three-tier model
Tier 1 — Foundation literacy (all staff): Every employee should understand: which AI tools are approved; what kinds of data should never enter AI tools; that AI output requires human review and is the users professional responsibility; and how to report AI-related concerns. This is the baseline — equivalent to WHS induction or data security awareness training.
Tier 2 — Applied skills (role-specific): Teams using AI daily need: how to prompt effectively for their use cases; how to critically evaluate AI output for accuracy and bias; what peer review is needed before acting on AI output; and how AI output should be disclosed when relevant to clients or regulators. What a finance team needs differs from what a legal team needs.
Tier 3 — Specialist capability (builders, procurers, governors): Those buying AI tools, building systems or governing AI risk need deeper capability: conducting AI risk assessments; evaluating vendor AI governance claims; designing human oversight mechanisms; monitoring AI performance over time; and managing AI-related incidents.
AI6 makes this a governance obligation
AI6 Practice 1 requires a named executive with genuine capability to discharge AI accountability. Practice 3 requires risk professionals who understand AI-specific risks. Practice 6 requires reviewers who can meaningfully oversee AI outputs. An organisation deploying AI at scale without building corresponding capability has a control gap, not just a training gap.
Free resources available now
The National AI Centre provides AI6 implementation guidance, AI policy templates, AI system register templates, and an AI screening tool — all free at industry.gov.au. TAFE NSWs AI microskill course is available through the million-scholarship program. These are a starting point, not a substitute for role-specific, workflow-embedded training.
