Can new job creation outweigh displacement, and what can support prove?

What the evidence shows

Three effects, no common measure

Displacement, augmentation and creation can occur together. Available Australian evidence measures them differently, so it cannot yet rank creation as the largest effect.

Augmentation is a competing force

Australian official analysis emphasises upskilling, redeployment and role evolution. International job-ad evidence also points to rapid skill change in AI-exposed work.

Barriers make targeted support relevant

Small firms report gaps in skills, time, security, finance and implementation. Government and intergovernmental plans respond to those barriers, supporting the case for testing targeted help.

Policy instruments are not outcomes

A programme can exist and improve access without proving that creation becomes the largest labour-market effect or that the resulting work provides viable income, survival and protection.

What these verdicts do not mean

• AI-enabled creation is small or absent
• displacement is proven to exceed creation
• targeted support is unnecessary
• existing support is sufficient to produce dominant, higher-quality creation
• all ages, regions and occupations face the same effects

Where this argument could be wrong

The H2 verdict could be too conservative if official statistics miss substantial viable new businesses, independent work and employee-to-owner transitions. The necessity finding could be too strong if inexpensive tools diffuse broadly without targeted intervention. Sufficiency could clear if a well-evaluated programme demonstrates viable work at scale.

Bottom line

Creation cannot yet be ranked above displacement and augmentation. Evidence partly supports targeted help as important for broad access, but neither strict necessity nor sufficiency is demonstrated. Support should therefore be treated as a test with pre-declared outcomes, not as proof that the hypothesis is true.

Selected source notes

These links support the bounded public extract. Source inclusion does not imply endorsement of the report's synthesis.

• Jobs and Skills Australia — Generative AI Capacity StudyCurrent Australian entry-level roles, role change and displacement limits.
• Australian Bureau of Statistics — Counts of Australian BusinessesBusiness counts and entries do not establish viable AI-caused work.
• PwC — 2025 Global AI Jobs BarometerInternational job-ad evidence on employment and skill change; not an official Australian causal estimate.
• OECD — Empowering SMEs in the age of AISME adoption barriers from a non-representative multi-country survey.
• Australian Government — National AI PlanPolicy intent and public capability roles; not realised labour outcomes.

Suggested citation