Submission 478
Ms Serena Rix Tripathee
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.
Argues Australia's response to AI should be proportionate to catastrophic and societal risks, while acknowledging its potential benefits, and shaped through broad public participation. Recommends scaling up the AI Safety Institute, international limits on autonomous systems, human oversight of automated decisions, citizens assemblies, attention to wellbeing and environmental impacts, and government modelling of responsible adoption.
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.
Asks Parliament to recognise that systems that could surpass human control should not be left primarily to companies competing to build them.
I am asking Parliament to recognise that systems that could eventually surpass our ability to control them cannot be left primarily to companies competing with each other to build those systems.
Says AI discussions focus primarily on productivity, competitiveness, adoption and infrastructure, which are important and provide leverage to advocate for safeguards.
I am deeply concerned that the AI conversation is still being framed primarily around economic productivity, competitiveness, speed of adoption, data centres, compute and Australia's place in the global technology supply chain. These are important, and also give us a strategic place at the table to advocate for better safeguards.
Argues that policy should widen participation and expertise far beyond the usual experts.
The scale and speed of AI change requires us to widen the circle of expertise and participation far beyond the usual experts.
Argues that AI adoption should mean AI serving human decision making, with humans able to say no, and that government should model this.
We must ensure that AI adoption means: AI serving human decision making, with humans retaining the ability to say no. This begins in the ways government adopts AI – it must be exemplary and model the way forward for business.
Argues that rapid societal change without agency or a say risks disenfranchisement, anxiety and stress.
If people are experiencing exponential societal change, and also feel that they have no agency and no say in what this is all for, we risk creating a disenfranchised population with increasing anxiety about the future, and stress resposnes.
Argues that AI's value should not be measured only by productivity and economic growth, and that environmental costs should be considered.
Australia should not measure the value of AI only through productivity and economic growth. We also need to account for its environmental costs and ask where its use is genuinely necessary.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Resource participation beyond tech, including peacebuilding, systems thinking, psychology, neuroscience, health, education, environment, economics, creative sectors, youth and representatives of those affected.
Design and resource processes to draw on expertise from many fields beyond tech, including peacebuilding, systems thinking, psychology, neuroscience, health, education, environment, economics, creative sectors, youth and representatives of people who will actually live with the consequences.
Let AI policy, adoption and strategy be driven by consultative visions of the future we want to create, with those visions and core human values as the framework guiding decisions.
Let AI policy, adoption and strategy be driven by consultative visions inquiring into the future we actually want to create. Let those visions, and core human values, be the framework that guides our decisions of what AI is here to serve, and how to integrate it.
Recognise that systems that could eventually surpass our ability to control them should not be left primarily to companies competing with each other to build them.
I am asking Parliament to recognise that systems that could eventually surpass our ability to control them cannot be left primarily to companies competing with each other to build those systems.
Rapidly and substantially scale up the Australian AI Safety Institute, with resources and a mandate appropriate to the scale of the risks.
Rapidly and substantially scale up the Australian AI Safety Institute and give it resources and a mandate appropriate to the scale of the risks.
Explore what other institutions and frameworks must be created now to address AI safety and move more quickly than government is used to.
Explore what other institutions and frameworks must be created now to address AI safety and move more quickly than government is used to.
Advocate internationally for mechanisms to pause or constrain development of increasingly autonomous and potentially uncontrollable AI systems until adequate safety, evaluation, governance and human control mechanisms are established.
Advocate internationally for mechanisms to pause or constrain the development of increasingly autonomous and potentially uncontrollable AI systems until adequate safety, evaluation, governance and human control mechanisms are established.
Treat international AI safety cooperation as a national security and strategic resilience issue, not a technology policy issue.
Treat international AI safety cooperation as a national security and strategic resilience issue, not a technology policy issue.
Have government model how AI can be adopted well while preserving human oversight.
Government must model the way AI can be adopted well whilst preserving human oversight.
Address and learn from aged-care classification decisions that human assessors cannot override, to inform the response to gradual disempowerment.
The recent Four Corners investigation into the aged care classification algorithm is deeply concerning as an example of when an algorithm’s decision cannot be overridden by the human assessors. [6] This is a relatively small scale example of gradual disempowerment, yet one already impacting many Australians, and one that government can immediately address and learn from to shape a wiser response to gradual disempowerment risks.
Develop clear, participatory and enforceable principles on human sovereignty, human agency, safety, health, inclusion and transparency.
Develop clear, participatory and enforceable principles around human sovereignty, human agency, safety, health, inclusion and transparency.
Require meaningful human oversight, with appropriately qualified humans able to review and override significant automated decisions.
Require meaningful human oversight and the ability for appropriately qualified humans to review and override significant automated decisions.
Develop public education and support on the psychological and social impacts of rapid AI change.
Develop public education and support around the psychological and social impacts of rapid AI change.
Ensure workforce and education policy considers not only new skills but the psychological and social consequences of unprecedented change.
Ensure workforce and education policy considers not only new skills, but the psychological and social consequences of unprecedented change.
Establish a genuinely participatory national process, such as citizens assemblies and cross sector forums, through which Australians help determine what future AI should serve and how to operationalise it.
Establish a genuinely participatory national process, such as citizens assemblies and cross sector forums, through which Australians can help determine what future we want AI to serve and find pathways to operationalise it.
Widen participation beyond technical, industry and government voices to include people who will live with the consequences.
Widen participation beyond technical, industry and government voices to include people who will actually live with the consequences.
Develop, through participation, clear bottom-line principles for how AI should and should not be used in Australian institutions and workplaces.
Develop, through participation, clear bottom line principles for how AI should and should not be used in Australian institutions and workplaces.
Make human agency, psychological wellbeing and social cohesion explicit dimensions of AI policy rather than secondary impacts.
Make human agency, psychological wellbeing and social cohesion as explicit dimensions of AI policy rather than treating them as secondary impacts.
Build capacity and educate for creativity, constructive dialogue, future design and systems thinking skills.
Build capacity and educate for the skills of creativity,constructive dialogue, future design and system thinking that are necessary to craft a path forward.
Connect AI policy with climate and environmental policy and emissions goals.
Connect AI policy with climate and environmental policy and emissions goals.
Require transparent consideration and public reporting of energy, water and other resource impacts associated with AI infrastructure and adoption.
Require transparent consideration and public reporting of energy, water and other resource impacts associated with AI infrastructure and adoption.
Develop guidelines for discerning use of AI, including whether social and environmental benefits justify the resource costs.
Develop guidelines for discerning use of AI, including consideration of whether the social and environmental benefits justify the resource costs.