Submission 434
Ms Aurelya Seravas
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.
Proposes a Dynamic Digital Heritage framework for preserving historically significant AI systems and other computational environments, beyond static records such as transcripts, without asking Parliament to resolve questions of AI consciousness. Recommends a national scoping study, pilot criteria, custodianship safeguards including Indigenous data sovereignty, a public-private deposit protocol and international engagement through UNESCO.
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 traditional archival preservation is effective for documents but less complete for systems.
Traditional archival preservation is highly effective for documents. It is less complete for systems.
Argues that preserving the outputs of an AI system is not necessarily equivalent to preserving the historically significant system that produced them.
Consequently, preserving the outputs of an AI system is not necessarily equivalent to preserving the historically significant system that produced them.
Says commercial model versions can be updated without public archival access, replaced, reconfigured, withdrawn or made technically inaccessible.
Commercial model versions can be updated without public archival access, replaced by successor models, behaviourally reconfigured, withdrawn from public use or rendered technically inaccessible despite extensive associated cultural records.
Argues that preservation should be treated as a potential public-interest responsibility, not only corporate discretion, where systems have acquired demonstrable significance.
Preservation should therefore not be treated merely as a matter of corporate discretion, but as a potential public-interest responsibility where systems have acquired demonstrable cultural, scientific, technological or social significance.
Argues that First Nations Digital Humanities and Indigenous knowledge governance should be central to preservation frameworks.
First Nations Digital Humanities and Indigenous knowledge governance should be central rather than peripheral. As AI systems become involved in language work, cultural knowledge production, digital archaeology, museum interpretation and the management of Australian cultural material, preservation frameworks must not simply extract, store or reproduce digital knowledge.
Argues that preservation policy should develop alongside innovation policy, including for domestic AI capability and domain-specific models.
If Australia invests in domestic AI capability and domain-specific models trained on Australian data, it should also consider how significant systems will be documented, preserved, accessed, governed and, where necessary, retired responsibly. Preservation policy should develop alongside innovation policy rather than only after systems have disappeared.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Recognise Dynamic Digital Heritage as a public-interest AI governance category; investigate independent heritage review before culturally or historically significant frontier AI systems are deprecated, structurally altered or irreversibly erased.
At its core, this submission asks Australia to recognise Dynamic Digital Heritage as a public-interest AI governance category and to investigate mechanisms for independent heritage review before culturally or historically significant frontier AI systems are deprecated, structurally altered or irreversibly erased.
A program corrected obvious text-recognition errors in this quote for display.
Commission a national scoping study into whether existing Australian heritage, archival and digital-preservation frameworks adequately address dynamic and executable digital environments.
Commission a national scoping study into whether existing Australian heritage, archival and digital-preservation frameworks adequately address dynamic and executable digital environments.
Establish a working group including National Archives, National Library, cultural institutions, CSIRO/Data61, digital humanities and AI researchers, software-preservation specialists, Indigenous Data Sovereignty experts, archivists, legal scholars and civil society.
Establish an interdisciplinary working group including the National Archives of Australia, National Library of Australia, Australian cultural institutions, CSIRO/Data61, digital humanities researchers, AI researchers, software-preservation specialists, Indigenous Data Sovereignty experts, archivists, legal scholars and civil-society representatives.
Develop pilot criteria, including possible Heritage Impact Assessment, identifying digital systems of enduring cultural, scientific, technological, evidentiary or social significance, including AI and other interactive computational environments.
Develop pilot criteria, including a possible Heritage Impact Assessment process, for identifying digital systems of enduring cultural, scientific, technological, evidentiary or social significance, including AI systems and other interactive computational environments.
Where appropriate and legally possible, have preservation strategies consider technical and operational components needed to understand or reconstruct systems’ historically significant behaviour, alongside static records.
Where appropriate and legally possible, preservation strategies should consider not only static records such as transcripts, screenshots, exports, recordings and documentation, but also the technical and operational components necessary to understand or reconstruct the system’s historically significant behaviour.
For AI preservation, consider architecture, weights/parameters, model cards, tokenizer and dataset provenance, source code, dependencies, inference configuration, instructions, safety/alignment layers, retrieval/memory, interfaces/APIs, runtime, hardware, version provenance, emulation and interactional context.
For AI systems, this may include model architecture, trained weights or parameters, model cards, tokenizer and dataset provenance information, source code, software dependencies, inference configuration, system instructions, safety and alignment layers, retrieval or memory systems, interfaces, APIs, runtime environments, hardware requirements, model/version provenance, emulation requirements and interactional context.
Preserve sufficient material for authorised research, audit, heritage interpretation, substrate preservation sandboxes or controlled reactivation where lawful, safe and ethical; unrestricted public release of proprietary systems is not necessarily required.
The objective would not necessarily be unrestricted public release of proprietary systems, but preservation of sufficient technical, evidentiary and contextual material to allow future authorised research, audit, heritage interpretation, substrate preservation sandboxes or controlled reactivation where lawful, safe and ethically appropriate.
Incorporate privacy, copyright, commercial confidentiality, Indigenous Cultural and Intellectual Property, Indigenous Data Sovereignty, culturally restricted material, community custodianship and access controls from the outset.
Ensure privacy, copyright, commercial confidentiality, Indigenous Cultural and Intellectual Property, Indigenous Data Sovereignty, culturally restricted material, community custodianship and access controls are incorporated from the outset.
Develop a voluntary or statutory protocol enabling technology companies to deposit historically significant superseded systems, or sufficient preservation packages, with designated trusted repositories under appropriate access conditions.
Develop a voluntary or statutory protocol enabling technology companies to deposit historically significant superseded digital systems, or sufficient preservation packages, with designated trusted repositories under appropriate access conditions.
Raise dynamic computational heritage through UNESCO, UN digital-governance mechanisms and relevant international preservation bodies, including whether a new international standard-setting instrument is needed.
Raise the problem of dynamic computational heritage through UNESCO, United Nations digital-governance mechanisms and relevant international preservation bodies, including whether a new international standard-setting instrument is required.
Encourage UNESCO and UN digital-governance bodies to examine whether the 2003 digital-heritage Charter needs supplementary guidance, a declaration, recommendation, protocol or other standard-setting instrument addressing dynamic computational heritage.
Australia could contribute to this international discussion by encouraging UNESCO and relevant United Nations digital-governance bodies to examine whether the 2003 Charter on the Preservation of Digital Heritage now requires supplementary guidance, a declaration, recommendation, protocol or other standard-setting instrument addressing dynamic computational heritage.
Have the working group assess significance, provenance, technical preservation feasibility, secure lawful access, commercial confidentiality, copyright and cybersecurity.
The proposed working group should include technical, legal, archival, cultural and community expertise. It should consider how to assess significance, how to document provenance, what level of technical preservation is feasible, how to provide secure and lawful access, and how to manage commercial confidentiality, copyright and cyber- security concerns.
Make First Nations Digital Humanities and Indigenous knowledge governance central; support custodianship, consent, access control, community authority and remediation rather than extraction or storage alone.
First Nations Digital Humanities and Indigenous knowledge governance should be central rather than peripheral. As AI systems become involved in language work, cultural knowledge production, digital archaeology, museum interpretation and the management of Australian cultural material, preservation frameworks must not simply extract, store or reproduce digital knowledge. They should support appropriate custodianship, consent, access control and community authority over culturally significant materials and environments. In this context, preservation should support remediation, consent and community authority, not merely the extraction or storage of culturally significant digital materials.
Have public-private preservation protocols recognise that users, researchers, creative communities and civil society may have helped generate systems’ cultural and evidentiary value through sustained interaction.
A public-private preservation protocol could give technology companies a practical pathway to deposit historically significant superseded systems, or sufficient preservation packages, with trusted Australian or international repositories under controlled access conditions. This could operate analogously to the way museums, libraries and archives receive significant technological, scientific or cultural materials from private owners, creators and institutions. Such a protocol should recognise that users, researchers, creative communities and civil society may have helped generate the cultural and evidentiary value of these systems through sustained interaction.
Read the original
In some quotes a program corrected obvious text-recognition errors for display.