AI Inquiry Atlas

An unofficial, independent guide to the public records of the Joint Select Committee on Artificial Intelligence. Not connected with the Parliament of Australia or the committee. Its makers also made a submission (no. 481; see About). Summaries and scores are written by AI and can be wrong: check the original before relying on them.

Submission 68

RMIT Enterprise AI and Data Analytics Research Hub

Submitter group: Universities.

An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.

Open in the interactive guide

Summary

Read the original submission for the full text.

Describes SME AI-adoption barriers using a single RMIT–Wyndham City Council program, without establishing national prevalence or causal productivity effects. Recommends evaluated local partnerships, staged implementation support, co-investment, practical guidance, outcome measurement, adviser capacity and student placements, worker consultation and job-quality monitoring, and a shared library with contributions required from publicly co-funded projects.

Key points

Quotes are the submitter’s own words. Text taken from documents, including scanned pages read by text-recognition software, can differ from the PDF. Check the original. Listed in the same order as on the submission’s page in the interactive guide.

  1. Says that program experience shows a gap between interest in AI and organisational implementation capability, including among participants who already use AI daily.

    Program experience highlights a gap between interest in AI and organisational implementation capability. Survey evidence indicates this gap runs through individuals as well as between them: participants who already use AI daily in their own work still default to task-level experimentation rather than organisational rollout.

  2. Argues that the barriers to SMEs adopting AI are complex and interconnected.

    The barriers to SMEs adopting AI are complex and interconnected.

  3. Argues that scoping capability is a precondition for adoption, reporting that participating businesses often could not name a data owner or specify success measures unaided.

    Scoping capability is a precondition for adoption, not a formality. Participating businesses frequently could not, without assistance, nominate a data owner, specify success measures or assemble a set of test cases.

  4. Argues that trusted place-based intermediaries can connect businesses with technical, research, governance and workforce expertise.

    Trusted place-based intermediaries can connect businesses with technical, research, governance and workforce expertise.

  5. Argues that workforce capability and worker consultation are part of implementation rather than a separate skills question.

    Workforce capability and worker consultation are part of implementation, not a separate skills question, and should be funded and monitored as such.

  6. States that its evidence comes from a single place-based program and is not a representative survey of Australian SMEs.

    The observations in this submission are practice-based evidence from a single place-based program. They are not a statistically representative survey of Australian SMEs and do not establish national prevalence or causal productivity effects.

Recommendations

Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.

  1. Test and adapt wider rollout of the Wyndham approach to different communities, including regional settings.

    Wider rollout should test and adapt this approach to different communities, including regional settings, rather than assume that the same model will transfer unchanged.

  2. Target local AI-adoption delivery at evidenced access gaps, including smaller and lower-readiness firms.

    Local delivery should target evidenced gaps in access, including smaller and lower-readiness firms.

  3. Have AI skills policy cover general literacy, role-specific skills and organisational implementation, including problem definition, vendor assessment, governance, work redesign and benefit measurement.

    AI skills policy should cover general literacy, role-specific capability and organisational implementation capability. Managers need to define problems, assess vendors, govern use, redesign work and measure benefits.

  4. Strengthen and connect existing support, including AI Adopt Centres, through evaluated partnerships among universities, local governments and industry organisations, including mapping access gaps, agreeing referral responsibilities and funding local implementation support where a gap is demonstrated.

    The Australian Government should strengthen and connect existing support, including AI Adopt Centres [5], through evaluated partnerships among universities, local governments and industry organisations. This should include mapping geographic and sector access gaps, agreeing referral responsibilities, and funding additional local implementation support where a gap is demonstrated.

  5. Include local-government boundary effects in access mapping.

    Access mapping should include local-government boundary effects, since businesses trade across boundaries and self-exclude quietly at the awareness stage.

  6. Make any funding or delivery-partner selection open and contestable, with transparent criteria and evaluation.

    Any funding or delivery-partner selection should be open and contestable, with transparent criteria and evaluation.

  7. Fund diagnostics, readiness, preparation, pilots, integration, workforce development and evaluation; cover handover, maintenance ownership and follow-up; condition deeper investment on proportionate readiness checks, business commitment and a credible pathway to value; separately fund free vendor-neutral advice and assisted problem definition, data ownership, success measures and test cases.

    Public programs should connect low-barrier diagnostics with readiness, preparation, pilots, integration, workforce development and evaluation. Post-pilot support should address handover, maintenance ownership and follow-up. Deeper investment should depend on proportionate readiness checks, business commitment and a credible pathway to value. Funding should treat the free, vendor-neutral advisory tier as a distinct fundable element and should cover assisted scoping (e.g. defining the problem, nominating a data owner, setting success measures and assembling test cases) as eligible activity.

  8. Provide matched or graduated co-investment for SMEs and family businesses according to size and demonstrated need, covering preparation, implementation, integration and workforce learning; require bounded scope, ownership, success measures, transparent selection, conflict controls and evidence of additional useful activity.

    Government should provide matched or graduated co-investment for SMEs and family businesses, calibrated to business size and demonstrated need. Eligible costs should cover preparation, implementation, integration and workforce learning. Require bounded scope, ownership, success measures, transparent selection and conflict-of-interest controls, with evidence that public support enables useful activity that would not otherwise proceed.

  9. Fund intermediaries to apply National AI Centre guidance through sector examples and proportionate use-case support prioritising risk screening, privacy, vendor due diligence, human oversight, incident response, workforce consultation and specialist referral where needed.

    The National AI Centre already provides adoption guidance and supporting templates [6]. Fund trusted intermediaries to help SMEs apply these resources through sector examples and proportionate, use- case-specific support. Priorities include risk screening, privacy, vendor due diligence, human oversight, incident response and workforce consultation, with specialist referral where needed.

  10. Provide a proportionate variant of existing AI guidance for firms where one person is owner, data custodian and approver.

    Additionally, existing guidance generally assumes separated organisational roles; a proportionate variant is needed for firms where one person is owner, data custodian and approver, and this variant does not presently exist.

  11. Require pre-pilot baselines and proportionate follow-up; distinguish participation, functionality, continued use and business and workforce effects; record implementation and operating costs, service quality, errors, incidents and stopping reasons; report progression denominators, separate self-reported and independently checked results, and use comparisons where feasible for causal claims.

    Require pre-pilot baselines and proportionate follow-up after handover. Distinguish participation, delivered functionality, continued use and measured business and workforce effects. Record implementation and operating costs, service quality, errors, incidents and reasons for stopping. Report progression denominators and distinguish self-reported from independently checked results; use comparison evidence where feasible for causal claims.

  12. Fund adviser capability and light accreditation, shared methods, scoping templates and reference architectures, and student placements in SME implementation under academic and industry supervision.

    Government should fund the supply side directly: capability development and light accreditation for AI adoption advisers; shared methods, scoping templates and reference architectures so that each intermediary is not building from scratch; and supervised work-integrated-learning placements that put students into SME implementation work under academic and industry supervision.

  13. Condition public co-funding on proportionate consultation with affected staff before deployment, paid training time and follow-up reports of role changes, workload and attributable redundancies; keep obligations simple for very small firms and demonstrate that funded adoption did not degrade job quality.

    Where public money co-funds AI implementation, worker consultation should be a condition of funding rather than a matter of good practice. Recipients should be required to consult affected staff before deployment on a proportionate basis, allow paid time for training, and report role changes, workload effects and any redundancies attributable to the funded work at follow-up. For very small firms this should be a short, simple obligation rather than a formal process. Publicly funded adoption should be able to demonstrate that it did not degrade job quality, and we are not aware of any Australian adoption program in this space that collects the data that would show either way.

  14. Fund an openly licensed national library of de-identified cases, scoping templates, evaluation instruments and reference architectures, maintained by a national body, with contributions required as a condition of public co-investment.

    Government should fund a shared, openly licensed library of de-identified cases, scoping templates, evaluation instruments and reference architectures, maintained by a national body and contributed to as a condition of public co-investment.

Read the original

Open in the interactive guide