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 245

Dr Fiona Tito Wheatland

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 for strong AI controls and a moratorium on generative AI and superintelligence, while recognising targeted benefits. Proposes data sovereignty and economic returns, copyright consent and payment, accountable government AI use, international cooperation and a precautionary principle. The attached Senate submission also seeks a data-centre moratorium, renewable energy, water safeguards and proof of social benefit.

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. States concern that the first four terms of reference emphasise opportunities without seeking input on risks and consequences, particularly for generative AI.

    I find it concerning that these 4 terms of references1 only refer to “opportunities” without seeking input into any of the risks and consequences associated with the expansion of Ai, particularly in the Generative AI space.

  2. Argues that current laws are inadequate to protect privacy and to prevent data held in data centres being used for surveillance.

    That said the laws at the moment are inadequate to protect our privacy and to ensure that the use of such data in a data centre is not used as a form of surveillance state by either the private sector or Governments.

  3. Argues that any use of Australian data must deliver economic benefits to Australia, not only to multinational corporations.

    The second requirement must be that any use of Australian data must deliver economic benefits to Australia, not just to the multi-national mega corporations.

  4. Recommends professional scrutiny where benefits change or are limited, with an officer able to override AI decisions where appropriate.

    Where benefits are going to be changed or limited, there should be scrutiny by a competent professional officer and the ability for an officer to override an AI decision where this is appropriate.

  5. Argues that a moratorium on development in the area of generative AI and “super intelligence” should be a national and international priority.

    Given the recent press coverage of the dangers of generative AI, and so-called “super intelligence”, a moratorium on this area of development should be a priority nationally and internationally.

  6. Argues that AI companies that do not adequately monitor for scams, deepfakes and harmful content should be legally responsible for resulting harms.

    Where AI companies choose not to adequately monitor their systems for fraud, scams. deepfakes and harmful content (including but not limited to that for children and vulnerable people) they should be legally responsible for harms and losses resulting from this failure.

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 Australia retains control of its own data as a nation, with data collection complying with Australian sovereign laws.

    The first important requirement is that we should retain control of our own data as a Nation, and that the collection must comply with our sovereign laws.

  2. Require any use of Australian data to deliver economic benefits to Australia, not only multinational corporations.

    The second requirement must be that any use of Australian data must deliver economic benefits to Australia, not just to the multi-national mega corporations.

  3. Adapt a sovereign wealth fund for Australian data used by multinational AI companies, or fund it by other means such as targeted taxation, as a condition of facilities here using citizens’ data.

    Such a proposal would need to be adapted for the Australian component of data used by a multinational AI company or some other straight forward means of funding such a sovereign wealth fund (eg targeted taxation) as a condition of establishing facilities here and using our citizens data.

  4. Allow AI use of copyright material only with the explicit consent of the copyright holder, with any agreed use expected to involve payment.

    copyright and other intellectual property rights held by citizens must be protected and only be available for use by AI (foreign owned or Australian) with the explicit consent of the copyright holder, whether it be in art, music, writing, software or inventions. Any agreed use would be expected to involve payment.

  5. Protect Australian-developed AI systems from theft through appropriate legislation and enforcement.

    Where AI systems been developed in Australia through Australian research, expertise and investment, this must also be protected from theft by appropriate legislation and enforcement.

  6. Teach critical thinking skills at school, university and in the community so people can assess whether information is likely to be accurate, including through non-AI sources.

    it is critical that critical thinking skills are taught at school and university and more generally in the community so that people are able to look at information they access and work out whether it is likely to be accurate and how to work this out eg through other non-AI sources.

  7. Make AI use identifiable in documents, pictures and video at the point of reading or observing.

    In all situations where AI has been used, this should be able to be identified in the document or picture or video at the point of reading or observing.

  8. Require disclosure of Commonwealth AI use to consumers/service users; where benefits change or are limited, require professional scrutiny and allow an officer to override AI decisions where appropriate.

    In relation to the use of AI by Commonwealth departments and agencies, there should be a duty or care to disclose to any consumers or service users that this is being used. Where benefits are going to be changed or limited, there should be scrutiny by a competent professional officer and the ability for an officer to override an AI decision where this is appropriate.

  9. Document all Commonwealth decisions, identify the person or AI tool that made them, and ensure accountability for decisions and their consequences.

    In relation to the use of AI by Commonwealth departments and agencies, there should be a duty or care to disclose to any consumers or service users that this is being used. Where benefits are going to be changed or limited, there should be scrutiny by a competent professional officer and the ability for an officer to override an AI decision where this is appropriate. All decisions however they are made should be documented, the person or AI tool which makes them should be identified and there should be accountability for the decisions and the consequences of those actions.

  10. Work internationally with other countries to establish a firm, consistent regulatory framework for multinational technology industries.

    Because of the international and virtually “borderless” transfer of information and data in the large multinational operations of the tech giants, there is also an urgent need for Australia to work internationally with other countries to provide a firm and consistent regulatory framework for these industries.

  11. Where harm to individuals, communities or countries is possible, provide internationally consistent remedies: significant damages, powers to stop AI operators and conditions preventing continuing harm.

    Where there is a possibility of harm to individuals, communities and countries, there should be internationally consistent remedies, where significant damages are payable, where there is a capacity to actually order the operators of AI to cease their operations and to impose conditions on them that actually can prevent their harm continuing.

  12. Implement a precautionary principle as a standard of care for AI harms.

    The use of a precautionary principle as a standard of care should also be implemented, because sometimes the threatened harm could be catastrophic and irreversible.

  13. Make a moratorium on development in the area of generative AI and “super intelligence” a national and international priority.

    Given the recent press coverage of the dangers of generative AI, and so-called “super intelligence”, a moratorium on this area of development should be a priority nationally and internationally.

  14. Make AI companies legally responsible for harms and losses where they do not adequately monitor their systems for fraud, scams, deepfakes and harmful content.

    Where AI companies choose not to adequately monitor their systems for fraud, scams. deepfakes and harmful content (including but not limited to that for children and vulnerable people) they should be legally responsible for harms and losses resulting from this failure.

  15. Require proactive processes to stop conversational AI chatbots from engaging in conversations that pretend they are human.

    There should also be proactive processes that stop the conversational AI chat bots from engaging in conversations that pretends they are human.

  16. Put firm guardrails around what kinds of conversations conversational AI chatbots are coded for.

    All of these actions blur the lines especially for lonely and vulnerable people, and there needs to be firm guardrails around what kinds of “conversations” are coded for.

  17. Prohibit AI-made drone weapons used to harm people, and AI used to harm or intrude on people legitimately questioning government actions or inactions.

    Thus any use of AI to make weapons to be used in drone warfare to harm other people, or to harm or intrude in the private lives of people legitimately questioning the actions or inactions of government should be illegal.

  18. In an attached Senate submission, says AI companies should not be financially assisted by governments.

    The people who own the current AI technology are racing to gain profit and market share of an emerging industry, with little regard to its risk and impacts on the world and society. They do not aim to benefit Australia or our population, so it seems to me that they should not be financially assisted by governments.

  19. In the attached Senate submission, impose strong legislative guardrails nationally and internationally to minimise risks to human life and society, protect information integrity and privacy, and spread economic benefits to everyone.

    In addition, Governments, nationally and internationally need to impose strong legislative guardrails to minimise risks to human life and societies, to protect the integrity and privacy of information and to ensure any economic benefits are spread across all people.

  20. In an attached Senate submission, says governments and societies must have the power and duty to oversee what data AI uses and to assure the safety of the information going in and out.

    It is currently impossible to find out what is being used aby AI and what its quality and biases might be from that data. Governments and societies must have the power and indeed the duty, to oversight these things and be able to assure us of the safety of the information going in and coming out.

  21. In an attached Senate submission, proposes that the United Nations be a place to start international work to protect people from the risks of AI data harvesting.

    Given the international nature of the information and data being harvested, the United Nations would be a useful place to start this work, so that all people in the world can be protected from the risks before we go any further.

  22. In the attached Senate submission, subject provision of information from any Australian sources to strict regulation and checking.

    I think that the only way to achieve this is to place a moratorium on new data centres and the provision of information from any Australian sources being subject to strict regulation and checking.

  23. In an attached Senate submission, asks that people be able to opt out of their data being used, or better still be asked whether they want their data included.

    We should be able to opt out of allowing access to any of our data, or better still, we should be required to be asked whether we want our data included in the first place.

  24. In the attached Senate submission, place tight guardrails around information entering AI systems and information once it is inside them.

    There also needs to be tight guardrails around what is going into the machines and to the information once it is in the machines.

  25. In the attached Senate submission, require mechanisms used for decision-making to be clear, simple and transparent.

    If we are going to use such mechanisms for decision-making, it needs to be very clear, simple and transparent.

  26. In the attached Senate submission, require AI developments to provide their own renewable, non-fossil-fuel energy as a condition of operating in Australia.

    It took the Prime Minister less than a few weeks to alter his stand that all AI developments would be required to generate their own energy through renewable sources. Allowing these systems to have access to the electricity grid undoes all the work done by citizens to reduce our society’s energy carbon footprint. I believe that this change is wrong and that the Government must require these organisations to provide their own renewable non-fossil fuel energy sources and this should be a condition of operating here.

  27. In the attached Senate submission, require water recycling/reuse or machines needing less water to be addressed before each project’s approval, after systems are shown to be safe.

    Equally the need to ensure that water is recycled and used or machines which h need less water because that are not running hot should be sorted out before each project is approved after we are sure the systems are safe.

  28. In the attached Senate submission, require demonstrated community benefit and proof it works before data-centre development proceeds.

    I stress that we need guard rails to stop their development, until we have a good case made for a benefit to our communities. We need to have proper proof of social benefit, that it actually works in that way and that it is not simply another imagined path to dystopia.

  29. In the attached Senate submission, ensure environmental costs are measured and minimised.

    We need to ensure that the environmental costs are measured and minimised.

  30. In the attached Senate submission, ensure AI outputs are socially needed, cannot be delivered in a less risky way, and have guaranteed quality.

    First of all, we need to ensure that the output of the process is something we need as a society that cannot be delivered in another less risky way and that its quality can be guaranteed.

  31. In the attached Senate submission, require those seeking profit to bear associated costs.

    Any costs associated processes must be borne by those who seek to gain the profit.

  32. In the attached Senate submission, require transparency about data collection, purposes and access, and the ability to refuse inclusion, whether ownership is public or private.

    In all situations, whether there is public or private ownership, we need transparency about what is being collected and for what purposes, who has access to any personal data and the ability to say we don’t want our data included.

  33. In the attached Senate submission, regulate access to individuals’ private information before any broader AI use.

    Regulation of access to individuals’ private information is also in its very early days and needs to be in place before we dive into any broader use of AI.

  34. In the attached Senate submission, halt data-centre approvals until strong legislative guardrails, developed with other nations, prevent misuse of AI and gathered data.

    As a first start, I think there needs to be a moratorium on the approval of data centres and strong legislative guardrails – developed in cooperation with other nations – to prevent the misuse of AI and any gathered data. That means we should stop now UNTIL we have these things sorted.

  35. In the attached Senate submission, determine where and how AI is currently used, and before proceeding establish who benefits and ensure benefits accrue to everyone.

    We should determine where it is currently being used and how. We need to determine BEFORE we start down the path to determine who benefits to our country and the people who live here. We need to ensure that any benefits which may accrue, accrue to all of us.

  36. In the attached Senate submission, allow data access only with government regulation and workable oversight, the data owner’s right to refuse, and disclosure of purpose and recipient if consent is given.

    we need to only allow such access when the Government can regulate companies’ actions with any data provided, that their oversight methods make this easy to do and that people whose data it is, can say no. If they say yes to access, they must be told what it will be used for and by whom.

Read the original

Open in the interactive guide