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 237

Deputy

Submitter group: Companies.

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.

Deputy, a workforce management software provider, describes deploying AI in shift-based workplaces and argues for AI-assisted, human-responsible use. It says existing laws apply and favours practical guidance before an AI-specific rulebook, while recommending consultation requirements, indirect-bias testing, regional adoption support and recognition of frontline shift workers in future-of-work policy.

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. Argues that the main SME adoption barrier is the distance between a generic promise and a safe, measurable use case.

    The main adoption barrier for small and medium businesses is not a lack of interest in AI. It is the distance between a generic promise and a safe, measurable use case inside a real business.

  2. Argues that frontline workers do not resist AI but that too much AI adoption in shift-based work happens to people rather than with them.

    Where it is introduced around them, without explanation, consultation or a visible benefit to their own working day, trust falls away. The problem is not that frontline workers resist AI. It is that too much AI adoption in shift-based work is currently happening to people, rather than with them.

  3. Argues that existing workplace, privacy, consumer and anti-discrimination laws already apply to AI and that what is missing is operational clarity.

    We do not think Australia's existing laws are inadequate in principle. Workplace, privacy, consumer, safety and anti-discrimination law already apply to AI-influenced decisions and should continue to. What we see missing, particularly for small and medium businesses, is operational clarity

  4. Argues that sovereign capability includes Australian application companies, not only foundation models.

    Owning or hosting a foundation model is one layer of sovereign capability, but it is not the only one that matters economically. Australia also needs domestic companies that can turn local sector knowledge, trusted data practices and product expertise into valuable applications

  5. States a governing principle that people use AI to decide faster but remain accountable for outcomes.

    AI-assisted, human-responsible. Technology should help people make faster, better-informed decisions. However, a person should remain accountable for the outcome.

  6. Recommends testing AI-assisted rostering and other automated HR decisions for indirect bias, even without protected attributes as inputs.

    require AI-assisted rostering and other automated HR decisions to be tested for indirect bias against groups such as carers, particular age groups and genders, even where no protected attribute is used as an input.

Recommendations

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

  1. Treat operational AI for SMEs as a productivity and resilience agenda, measuring success across manager time, compliance, service continuity, employee experience and business performance, not labour cost alone.

    Treat operational AI for SMEs as a productivity and resilience agenda, and measure success across manager time, compliance, service continuity, employee experience and business performance, not labour cost alone.

  2. Encourage businesses and government agencies using workplace AI to track usage depth and confirmation/override rates alongside adoption counts; prefer individually verified case evidence over aggregate estimates and label aggregate time savings as estimates.

    Encourage businesses and government agencies adopting workplace AI to measure success the way we try to: tracking depth of use and confirmation or override rates alongside adoption headcounts, preferring individually verified case-level evidence over aggregate estimates, and labelling any aggregate time-saving figure clearly as an estimate.

  3. Fund regional demonstrators, trusted local advisors, mobile-first tools and short, job-specific training so smaller and regional businesses can use AI safely without an in-house AI team.

    Fund regional demonstrators, trusted local advisors, mobile-first tools and short, job-specific training, so smaller and regional businesses can see relevant use cases and gain value safely without an in-house AI team.

  4. Support sector-led application development, cross-industry demonstrations and export pathways for trusted Australian operational software, and value applications and workflow expertise built on infrastructure, not only the infrastructure itself.

    Support sector-led application development, cross-industry demonstrations and export pathways for trusted Australian operational software, and value the applications and workflow expertise built on top of infrastructure, not only the infrastructure itself.

  5. Shift support from broad awareness campaigns to sector playbooks, tool-selection support, small pilots, prompting guides and measurable case studies for common workflows such as rostering, payroll, training and compliance.

    Shift support from broad AI awareness campaigns to practical adoption pathways: sector playbooks, tool-selection support, small pilots, prompting guides and measurable case studies for common workflows as rostering, payroll, training and compliance.

  6. Pair infrastructure investment with support for Australian application companies, research partnerships, talent pathways and commercialisation programs that help domain-specific AI move from pilot to trusted deployment.

    Pair infrastructure investment with support for Australian application companies, research partnerships, talent pathways and commercialisation programs that help domain-specific AI move from pilot to trusted deployment.

  7. Create clear, practical guidance for SMEs and providers on data minimisation, access controls, audit trails, lawful data use and privacy-preserving research partnerships.

    Create clear, practical guidance for SMEs and providers on data minimisation, access controls, audit trails, lawful data use and privacy-preserving research partnerships.

  8. Require consultation, explanation and training wherever workplace AI materially affects a person's pay, hours or working conditions, and judge adoption by improved capability and job quality, not output alone.

    Require practical consultation, explanation and training wherever workplace AI materially affects a person's pay, hours or working conditions, and judge successful adoption by improved capability and job quality, not output alone.

  9. Fund role-based AI capability building for owners, managers and frontline employees, as shift-based workers are often under-served by generic digital-skills programs.

    Fund role-based AI capability building for owners, managers and frontline employees, recognising that shift-based workers are often under-served by generic digital-skills programs.

  10. Encourage and support AI-assisted rostering that gives shift workers, not only managers, earlier visibility of their own hours and expected earnings.

    Encourage and support AI-assisted rostering that gives shift workers themselves, not only managers, earlier visibility of their own hours and expected earnings, recognising that a significant share of Australia's shift-based workforce relies on coordinating two or more part-time or casual jobs to reach a full-time equivalent income.

  11. Require Commonwealth agencies to publish practical operating standards for consequential AI use and share reusable implementation patterns for smaller public bodies and regulated sectors.

    Require Commonwealth agencies to publish practical operating standards for consequential AI use and share reusable implementation patterns that smaller public bodies and regulated sectors can adopt.

  12. Give practical guidance on what a regulator such as the Fair Work Ombudsman should expect small businesses to check, record or audit when using AI-assisted payroll tools, without new AI-specific law.

    What is less clear is what a regulator, or the Fair Work Ombudsman, should reasonably expect a small business to have checked, recorded or audited when it relies on an AI-assisted payroll tool, and we think that is exactly the kind of practical guidance the Committee could usefully recommend, without adding a new layer of AI-specific law.

  13. Before a duplicative AI-specific rulebook, establish proportionate sector-aware guidance on human accountability, notice/explanation, consultation/training, permission controls, auditability and accessible review/correction, targeted at genuine gaps shown by deployment evidence.

    Before creating a duplicative AI-specific rulebook, establish proportionate, sector-aware guidance covering human accountability, notice and explanation, consultation and training, permission controls, auditability and accessible review and correction, targeted at genuine gaps revealed by real deployment evidence.

  14. Require AI used in recruitment, performance management and payroll to meet the same transparency, contestability and accountability standards as a human decision-maker.

    Require AI used in recruitment, performance management and payroll to meet the same standards of transparency, contestability and accountability expected of a human decision-maker

  15. Require AI-assisted rostering and other automated HR decisions to be tested for indirect bias against groups such as carers, particular age groups and genders, even where no protected attribute is used as an input.

    require AI-assisted rostering and other automated HR decisions to be tested for indirect bias against groups such as carers, particular age groups and genders, even where no protected attribute is used as an input.

  16. Preserve a fair and workable copyright framework for Australian creators and rights-holders while ensuring small businesses can use legitimate AI products with confidence.

    Preserve a fair and workable copyright framework for Australian creators and rights-holders while ensuring small businesses can use legitimate AI products with confidence.

  17. Apply controls proportionately to task consequences: strong access controls, logging, testing, escalation pathways and mandatory human review for AI materially affecting employment, pay or safety.

    Apply risk controls proportionately to the consequence of the task: strong access controls, logging, testing, escalation pathways and mandatory human review for any AI materially affecting a person's employment, pay or safety.

  18. Encourage secure-by-design procurement and plain-English vendor assurance for SMEs covering data handling, access controls, incident response and supply-chain transparency.

    Encourage secure-by-design procurement and plain-English vendor assurance for SMEs, covering data handling, access controls, incident response and supply-chain transparency, so that cyber resilience is built into everyday workplace AI products rather than added after the fact.

  19. Give the Australian AI Safety Institute and relevant regulators direct channels to hear from deployers, workers, SMEs and sector bodies, not only model developers.

    Give the Australian AI Safety Institute and relevant regulators direct channels to hear from deployers, workers, SMEs and sector bodies, not only model developers.

  20. Calibrate government guidance to consequence and context; as settings become more sensitive, ensure a person remains the final decision-maker before action.

    Government guidance should be calibrated to consequence and context, not applied as one uniform rule for every shift-based business, and the more sensitive the setting, the more essential it is that a person remains the final decision-maker before an action is taken.

  21. Treat the deskless, hourly and shift-based workforce as a distinct and significant part of the future-of-work agenda, not a subset of office-based, knowledge-work assumptions.

    Treat Australia's deskless, hourly and shift-based workforce as a distinct and significant part of the “future of work” agenda, not a subset of office-based, knowledge-work assumptions.

Read the original

Open in the interactive guide