Submission 538
Professor Victor Minichiello
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.
An academic’s individual submission proposes the Social Value Platform Development model, developed principally for social media, for testing rather than as a proven solution. It proposes independently evaluated public-interest outcomes alongside commercial viability in AI development, with community participation, human oversight, research access, public complaint/referral capacity and proportionate accountability when evidence conflicts with commercial decisions.
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 regulation after harm does not answer what kinds of technologies should be developed and what social outcomes design should support.
But regulation after harm does not answer a prior question: what kinds of technologies should be developed, and what social outcomes should their design be expected to support?
Argues that claims a technology creates social value should be open to examination rather than assumed.
The SVPD model begins with a simple proposition: claims that a technology creates social value should be capable of examination rather than assumed.
Argues that independent public-interest research can examine questions that may not align with commercial priorities.
Independent public-interest research performs a different function: it can examine questions that may not align directly with commercial priorities, including effects on user agency, differences between population groups, persistence of intended outcomes and unintended social consequences.
Proposes possible “comply and explain” responses to significant independent findings, proportionate to organisational scale, maturity and technological consequences.
One possible mechanism is a proportionate ‘comply and explain’ approach: where significant independent findings are produced, developers would state what action they have taken in response or explain why they have decided not to act. The form and burden of such accountability should be proportionate to the scale and maturity of the organization and the potential consequences of the technology.
Argues that a stronger public-interest approach to AI is not anti-innovation or anti-commercial.
A stronger public-interest approach to AI should not be understood as anti-innovation or anti-commercial. Australian developers need viable markets, investment, technical expertise, users and sustainable revenue models.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Examine social-outcome assessment alongside technical performance, safety, adoption, productivity and commercial success, particularly for consequential AI systems.
Social-value evaluation — whether Australia's developing AI frameworks should encourage assessment of social outcomes alongside technical performance, safety, adoption, productivity and commercial success, particularly for consequential AI-enabled systems.
Examine whether Australians have clear, coordinated ways to raise AI concerns and whether a public-facing concerns/referral function is needed.
Accessible public complaint and issue-reporting infrastructure — whether Australians have sufficiently clear and coordinated mechanisms for raising concerns about AI-enabled technologies, and whether a public-facing AI concerns and referral function is needed.
Examine whether funding and infrastructure support independent, longitudinal and interdisciplinary research into AI’s social consequences, including effects on children, young people and vulnerable populations.
Independent public-interest research — whether funding and research infrastructure are sufficient for independent, longitudinal and interdisciplinary investigation of the social consequences of AI, including effects on children, young people and vulnerable populations.
Examine whether regulators, researchers, consumer organisations, civil society and affected communities have capacity and evidence access to scrutinise consequential AI, including when independent findings conflict with commercial or product decisions.
Public-interest accountability infrastructure — whether regulators, researchers, consumer organisations, civil society and affected communities have sufficient capacity and access to evidence to scrutinise consequential AI systems, including when independent findings conflict with commercial or product decisions.
Examine opportunities for Australian developers, universities and communities to test alternative AI products/services, independently evaluating viability, user experience and social outcomes; consider strengthening this capability through Canadian and European collaboration.
Responsible Australian experimentation and innovation — whether policy provides sufficient opportunities for Australian developers, universities and communities to test alternative AI-enabled products and services while independently evaluating commercial viability, user experience and social outcomes, and whether collaboration with Canada and European partners could strengthen this capability.
Address values, accountability and purposes in AI design; require meaningful human oversight of consequential systems, with responsibility for risk review and intervention and preservation of human accountability.
Public-interest AI policy should therefore address not only identifiable harms, but also the values, accountability arrangements and purposes built into systems from the beginning (Floridi et al., 2018). For consequential AI systems, this should include meaningful human oversight, with clear responsibility for reviewing significant risks, intervening where necessary and ensuring that automated systems do not displace human accountability.
Extend evaluation beyond adoption, user numbers, engagement, technical performance, productivity and commercial return to examine socially beneficial outcomes.
Evaluation should therefore extend beyond adoption, user numbers, engagement, technical performance, productivity and commercial return. These measures remain important, but they do not necessarily show whether a technology is producing socially beneficial outcomes.
Make social-benefit claims explicit and independently examinable, particularly for systems affecting children, young people or populations less able to challenge decisions.
The key principle is that social-benefit claims should be explicit and independently examinable, particularly where systems affect children, young people or populations with less capacity to challenge decisions affecting them.
Consider independent, longitudinal, interdisciplinary research capacity with appropriate data access, privacy and minimisation, research independence/publication rights, transparent protocols, ethical oversight and affected-community participation.
Australia should consider whether its developing AI infrastructure provides sufficient capacity for independent, longitudinal and interdisciplinary research. This includes appropriate researcher access to data, privacy and data minimisation, research independence and publication rights, transparent protocols, ethical oversight and participation by affected communities.
Consider an identifiable AI concerns/referral function within or across existing institutions, providing entry, guidance and referral and identifying recurring/systemic issues for research, regulation or public reporting.
Australia should therefore consider a clearly identifiable public-facing AI concerns and referral function, located within an existing institution or coordinated across existing bodies. It could provide a visible entry point for concerns, guidance and referral, while also identifying recurring or systemic issues that warrant research, regulatory attention or public reporting.
Equip regulators and public bodies with technical, research and investigative expertise; give civil society, consumer groups and affected communities meaningful ways to contribute evidence to debate and regulation.
Regulators and other public bodies need sufficient technical, research and investigative expertise, while civil-society organisations, consumer groups and affected communities need meaningful ways to bring evidence into public debate and regulatory processes.
Support public-interest research infrastructure through competitive funding, longitudinal research and appropriate researcher data access.
Government can further support this infrastructure through competitive research funding, longitudinal research and appropriate mechanisms for researcher access to data.
Keep significant independent evidence about consequential AI, responses and decision reasons open to scrutiny, subject to legitimate privacy, security and commercial-confidentiality requirements.
Where significant independent evidence raises concerns about a consequential AI-enabled system, there should be mechanisms through which the evidence, the response to it and the reasons for important decisions remain open to scrutiny, subject to legitimate privacy, security and commercial-confidentiality requirements.
Consider proportionate “comply and explain”: developers respond to significant independent findings or explain inaction, with accountability burdens scaled to organisational size, maturity and technological consequences.
One possible mechanism is a proportionate ‘comply and explain’ approach: where significant independent findings are produced, developers would state what action they have taken in response or explain why they have decided not to act. The form and burden of such accountability should be proportionate to the scale and maturity of the organization and the potential consequences of the technology.
Provide affected people opportunities to identify and evaluate intended outcomes; include children’s interests and participation during design, recognise parents’ protective/advisory roles and young people’s evolving agency, and tailor participation to technology, population and consequences.
People affected by technologies should have appropriate opportunities to contribute to the identification and evaluation of intended outcomes. This is particularly important for children and young people. Parents may have legitimate protective and advisory roles, while young people also possess evolving agency and direct knowledge of their digital experiences. Child-rights-by-design approaches similarly emphasise that children's interests and participation should be considered during design rather than only after harms arise (Livingstone & Pothong, 2023). Participation should be matched to the technology, population and potential consequences under consideration.