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 179

Professor Lyria Bennett Moses

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.

Writes in a personal capacity on terms of reference (i) and (k). Argues existing law has significant gaps when applied to AI, using anti-discrimination law as an example, and that reform of discrimination, privacy and administrative law may work better than AI-specific regulation. Says some harms are not tied to AI as defined, and that this does not reduce the urgency of government action.

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 there are significant gaps in existing law when applied to AI systems.

    There are significant gaps in existing law when applied to AI systems.

  2. Says anti-discrimination laws are typically designed around an assumption that decision-makers are humans who might be prejudiced against protected groups.

    Anti-discrimination laws exist to protect against discrimination on protected grounds. Such laws are typically designed around an assumption that decision-makers are humans who might be prejudiced against protected groups.

  3. Argues that AI-specific regulation is less likely to resolve discrimination problems than discrimination-law reform, given the hybrid nature of much decision-making.

    There are thus good reasons to explore reform of discrimination law, including to ensure it better aligns with increasing reliance on AI systems in decision-making across sectors. Regulating AI specifically is less likely to be effective in resolving these problems given the hybrid nature of much decision-making.

  4. Says some harms, such as Robodebt and scam bots, arise from systems that would not meet the EU AI Act definition of an AI system.

    Robodebt adversely impacted vulnerable Australians despite the fact that the technical architecture was relatively unsophisticated. Similarly, even ordinary automated ‘bots’ can be used to enable frauds and scams. Neither would be an AI system under the definition of an AI system adopted by the European Union’s Artificial Intelligence Act in article 3(1)

  5. Argues that AI may not be the appropriate regulatory category for protecting Australians against harms from overreliance on systems.

    If legal requirements are to be put in place to protect Australians from harms associated with overreliance on systems, we should be careful before assuming that “artificial intelligence” is necessarily the appropriate category in all cases.

  6. States that the need for thoughtful law reform does not diminish the urgency of government action given the potential for serious harm.

    None of that diminishes the urgency of government action in light of the potential for serious harm.

Recommendations

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

  1. Ensure Australian laws are well adapted so that they apply appropriately, including where AI systems might cause harm or behave contrary to Australian values.

    It is important to ensure Australian laws are well adapted so that they apply appropriately, including in contexts where AI systems might cause harm or behave in ways contrary to Australian values.

    A program corrected obvious text-recognition errors in this quote for display.

  2. Undertake law reform thoughtfully, not assuming the problem is 'artificial intelligence' or that common AI definitions best target a concern, and always ask how laws are scoped.

    Law reform should be done thoughtfully rather than automatically assuming that the problem to be addressed is “artificial intelligence”. Where concerns relate to some features or behaviours of AI systems, it should not be assumed that the collection of features in common definitions is the best way to target the concern. How laws are scoped is an important question that should always be asked.

    A program corrected obvious text-recognition errors in this quote for display.

  3. Act urgently in light of the potential for serious harm, while pursuing thoughtful law reform.

    Law reform should be done thoughtfully rather than automatically assuming that the problem to be addressed is “artificial intelligence”. Where concerns relate to some features or behaviours of AI systems, it should not be assumed that the collection of features in common definitions is the best way to target the concern. How laws are scoped is an important question that should always be asked. 3. None of that diminishes the urgency of government action in light of the potential for serious harm.

    A program corrected obvious text-recognition errors in this quote for display.

  4. Support standards development, as proposed in Harm Ellens and colleagues’ submission to the committee.

    I also support the submission to this Committee made by Harm Ellens et al concerning support for standards development.

  5. Explore reform of discrimination law so it better fits AI-reliant decision-making.

    There are thus good reasons to explore reform of discrimination law, including to ensure it better aligns with increasing reliance on AI systems in decision-making across sectors.

  6. Treat reform of privacy, administrative and consumer protection law as similarly necessary areas of law reform.

    Similar points can be made about many other areas of necessary law reform including privacy law, administrative law, consumer protection law, and other relevant domains.

Read the original

In some quotes a program corrected obvious text-recognition errors for display.

Open in the interactive guide