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 124

Professor Alysia Blackham

Submitter group: Individuals.

Position labels are not shown for individual submitters.

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.

Argues AI and automated decision-making tools in the workplace risk bias and discrimination, and that Australian privacy and discrimination law have gaps. Recommends removing the small business and employee records exceptions, adopting EU-style transparency measures, a shifting burden of proof, and positive equality duties on employers.

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 automated decision-making tools at work carry risks of bias, discrimination and error.

    There are currently significant risks with adopting automated decision-making tools (including those using AI) in the workplace, including the risk of bias, discrimination and error.

  2. Argues that AI and automated decision-making tools raise considerable risks of bias and discrimination, which might be replicated at scale.

    The use of AI and other automated decision-making tools therefore raises considerable risks of bias and discrimination, and the risk that these in-built biases might be replicated at scale.

  3. Says Australian workers have few data protection rights, although AI tools are built on substantial quantities of data.

    AI tools are built on substantial quantities of data, yet Australian workers have few data protection rights.

  4. Recommends removing the Privacy Act's small business and employee records exceptions.

    The federal Parliament should remove the small business and employee records exceptions to the Privacy Act 1988 (Cth).

  5. Recommends a shifting burden of proof in federal discrimination law so employers must show their AI tool is not discriminatory.

    A shifting burden of proof should be adopted in federal discrimination law, so that employers need to show that their AI tool is not discriminatory.

  6. Argues that positive, proactive duties on employers are needed rather than reliance on individual complaints.

    Further, rather than relying on individuals to make a complaint, it is necessary to adopt positive, proactive duties on employers.

Recommendations

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

  1. Adopt rigorous privacy law like the GDPR as an important first step for Australian law reform.

    Adopting rigorous privacy law – like the GDPR – would be an important first step for Australian law reform.

  2. Require transparency so the data automated decision-making tools are based on, their use and their outcome are understandable, communicable and able to be effectively challenged.

    Transparency: so the data automated decision-making tools are based on, their use, and their outcome, is understandable, communicable, and able to be effectively challenged.

  3. Set substantive limits on when data can be collected and when automated decision-making tools can be used.

    Substantive limits on when data can be collected, and when automated decision-making tools can be used.

  4. Require oversight of automated decisions or outcomes by a human decision-maker ('human-in-command').

    Oversight of automated decisions or outcomes by a human decision-maker – an approach described as ‘human-in-command’.

  5. Require evaluation and consultation before AI decision-making tools are adopted, to check legitimate purpose, appropriateness and safeguards for individual rights.

    Evaluation and consultation prior to the adoption of AI decision-making tools, to ensure the technology is being used for a legitimate purpose, is appropriate for achieving that purpose, and includes appropriate safeguards for individual rights.

  6. Consider adopting EU-style measures to address gaps in Australian law on privacy, data protection, consultation and transparency.

    The federal Parliament should consider adopting similar measures to the EU to address the gaps in existing Australian laws and regulations relating to privacy, data protection, consultation and transparency.

  7. Remove the small business and employee records exceptions to the Privacy Act 1988 (Cth).

    The federal Parliament should remove the small business and employee records exceptions to the Privacy Act 1988 (Cth).

  8. Adopt a shifting burden of proof in federal discrimination law, so employers need to show their AI tool is not discriminatory.

    A shifting burden of proof should be adopted in federal discrimination law, so that employers need to show that their AI tool is not discriminatory.

  9. Extend positive equality duties to all protected characteristics under federal discrimination law.

    The federal Parliament should extend positive equality duties to all protected characteristics under federal discrimination law.

  10. Develop regulations ensuring positive equality duties specifically address the discrimination risks posed by AI tools.

    The federal Government should develop regulations that ensure positive equality duties specifically address the discrimination risks posed by AI tools.

  11. Impose positive equality duties requiring employers to consider and proportionately eliminate discrimination, advance equal opportunity, collect, analyse and publicise protected-characteristic data, and report data purposes, processing, use and quality controls.

    In particular, I have recommended21 the adoption of positive equality duties imposing obligations on employers that include requirements to: • give proper consideration and take proportionate action to eliminate discrimination and advance equality of opportunity;22 • collect, analyse and publicize data on the protected characteristics of those in the workplace;23 • report on what data is being collected, for what purpose, and how it is being processed and used, and data quality control measures adopted;

  12. Require reporting on adopted automated decision tools: operation, purpose, training, testing and organisational data, quality measures, data-review recency, workplace context and human oversight.

    report on what automated decision-making tools are being adopted, including: • how the technology operates, and for what purpose, • the training and testing and organisational data being deployed, data quality measures in place, and how recently data were reviewed or renewed, • how the system operates in the context of the specific workplace,24 • the role of human oversight in the process, and

  13. Require reporting on automated decision outcomes across the workforce and protected characteristics, policies addressing and eliminating discrimination, and consultation with workers and representatives on data collection and automated tool adoption and use.

    outcomes of automated decision-making tools across the workforce, including across protected characteristics; • adopt policies to demonstrate what is being done to address and eliminate discrimination; and • consult and engage with workers and worker representatives in the collection of data, and adoption and use of automated decision-making tools.

  14. Make positive equality duties enforceable by a statutory regulator such as the Australian Human Rights Commission with powers akin to s 47C, and by workers and their representatives.

    These duties should be enforceable by a statutory regulator, like the Australian Human Rights Commission, with powers akin to those in relation to the positive duty in s 47C. They should also be enforceable by workers and worker representatives, to ensure multiple checks on compliance.

Read the original

Open in the interactive guide