Submission 244
Dr Fan Yang and Dr Robbie Fordyce
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.
Summary
Read the original submission for the full text.
Two university researchers argue that AI productivity claims overlook data costs, outsourcing creates security and oversight risks, and Australian data workers lack protection. Recommend worker transparency, human decision-makers, opt-out rights, domestic infrastructure, research investment and co-design. Describe technological autonomy as independent control of critical technology stacks, without isolation or building everything from scratch.
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.
Argues that achieving organisational productivity from AI requires ongoing data investment, the cost of which is often not accounted for and may exceed that of maintaining a human workforce.
Achieving organisational productivity requires sustained and ongoing investment in data, the cost of which is often not accounted for and may exceed that of maintaining a human workforce.
Argues that outsourcing data work to third-party platforms moves organisational data across borders, outside the data protection legislation of the originating jurisdiction.
In the course of this data processing, organisational data, including confidential information, has already crossed jurisdictional borders and is consequently no longer protected by the data protection legislation of the originating jurisdiction.
Argues that Australian workers in the data annotation industry are in exploitative conditions and that current labour legislation is not well equipped to protect them.
An emerging, highly skilled workforce in Australia has become involved in the highly exploitative data annotation industry to train AI models, and our current labour legislation is not well equipped to protect this critical workforce.
Argues that employers should be transparent with workers about how AI is used at work and that workers should be able to opt out.
Employers must therefore undertake a mapping exercise to identify where AI is currently in use across their existing hardware and software, and must be transparent with workers about how it is being used. Workers must, in turn, be given the option to opt out of any engagement with AI, without being compelled to consent to its organisational use.
Argues that Australia is adopting AI within a framework built and hosted by American Big Tech and that the relationship requires urgent reassessment.
Australia is sleepwalking into AI adoption within a framework built and hosted by American Big Tech. This relationship with American Big Tech requires reassessment urgently
Argues that reliance on opaque proprietary systems is risky for public services, citing Robodebt as a lesson on offloading decisions to automated systems that are almost impossible to audit.
This risk is particularly acute in the provision of public services, where such systems increasingly underpin automated decisions affecting people’s social welfare and livelihoods. The Robodebt Scheme stands as a costly and consequential lesson on the risks of offloading human decision-making to automated systems that are, in practice, almost impossible to audit.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Develop evaluation metrics for assessing organisational productivity gains from AI.
Developing evaluation metrics for assessing organisational productivity gains: Achieving organisational productivity requires sustained investment in data, the cost of which is often overlooked and may exceed that of maintaining a human workforce.
Restrict outsourcing of core organisational functions and services, and invest in data localisation.
Restricting the outsourcing of core organisational functions and services and investing in data localisation:
Invest in organisational data control.
Restricting the outsourcing of core organisational functions and services and investing in data control:
Localise Australia's critical data workforce, develop long-term career pathways and legal protection for it, and support long-term job creation in the AI-driven economy.
Localising Australia’s critical data workforce and developing long-term career pathways and legal protection: An emerging, highly skilled Australian workforce has become involved in the exploitative AI data industry training AI models and our current labour legislation is not equipped to protect. Australia must support its long-term job creation in the AI-driven economy.
Require workers to communicate with, and be assessed by, human decision-makers.
Workers must always communicate with, and be assessed by, human decision-makers in a humanity centric workplace: Workers experience physical and mental strain from working with or for AI, alongside weakened workplace bargaining power.
Require employers to map where AI is in use and be transparent with workers about it; give workers the option to opt out without compelled consent; protect conscientious objectors from dismissal or Performance Improvement Plans for refusing to use AI.
Employers must therefore undertake a mapping exercise to identify where AI is currently in use across their existing hardware and software, and must be transparent with workers about how it is being used. Workers must, in turn, be given the option to opt out of any engagement with AI, without being compelled to consent to its organisational use. Protections must also be established for conscientious objectors to AI use, ensuring that workers cannot be dismissed or placed on a Performance Improvement Plan (PIP) for refusing to use AI.
Establish a Future of AI Made for All in Australia initiative, investing in technological autonomy and domestic infrastructure ownership through long-term economic strategy, research and development, and legislative reform.
Establishing a Future of AI Made for All in Australia initiative and towards technological autonomy: In alignment with the Future Made in Australia initiative, Australia must invest in building its own technological autonomy and domestic infrastructure ownership, requiring long- term vision across economic strategy, research and development, and legislative reform.
Pursue strategic technological autonomy through independent control, inspection, audit, modification and operation of critical technology stacks, without technological isolation or assuming all capabilities must be built from scratch.
Technological autonomy does not equate to technological isolation, which would in any case be impractical. It entails strategic autonomy: the capacity to control, inspect, audit, modify, and operate critical technology stacks independently. It would be a misconception to assume that achieving this requires building all such capabilities from scratch,
Provide stronger and more sustained government investment in university research on AI, including on Indigenous communities and culture, labour, ethics, economics, the environment, health, culture and creativity.
Stronger and more sustained investment in university research: As AI is a multidisciplinary field whose development and consumption impose significant impacts on both human societies and the environment, its development and innovation require government investment in research not only on AI and technology itself, but also Indigenous communities and culture, labour, ethics, economics, the environment, health, culture, and creativity.
Form a regulatory environment conducive to co-design, through close engagement among Australian businesses, universities and entities with expertise in Australia’s distinctive problems and systems.
Forming a regulatory environment conducive to co-design approaches in AI development. AI tools can only meaningfully address an organisation’s problems when such problems can be well understood by the organisation and by those developing the AI solution. A co-design approach is therefore essential, facilitated through close engagement among Australian businesses, universities, and entities with demonstrated expertise in Australia’s distinctive problems and systems.
Require sustained and substantive co-design engagement between businesses, researchers and target users when developing AI tooling.
A co-design approach is therefore needed in the development of AI tooling, requiring sustained and substantive engagement between business entities, researchers, and target users.