Submission 57
Association for Interdisciplinary Metaresearch and Open Science
Submitter group: Civil society.
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 that reliance on foreign AI models threatens the integrity and context-sensitivity of Australian research. Supports developing Australian AI models that are documented, inspectable, versioned and independently evaluable, and asks that the research community, including metaresearch experts, take a formal role in AI governance.
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 reliance on foreign AI models threatens the Australian research ecosystem.
Reliance on foreign AI models threatens the Australian research ecosystem
Argues that AI lowers the cost of producing fraudulent and low-quality research.
AI lowers the cost of producing fraudulent and low-quality research.
Argues that even declared AI use raises research integrity issues, especially with poorly understood offshore models lacking training and operational transparency.
Even when researchers fully declare their use of AI, that use still raises research quality and integrity issues, especially when relying on poorly understood offshore AI models that do not fully report how they were trained and how they operate.
Argues that there is no reason to expect a foreign AI vendor to prioritise Australian legal, clinical and cultural contexts.
there is no reason to expect a foreign AI vendor to prioritise Australian legal, clinical, and cultural contexts.
Says Australian AI models should be documented and inspectable, with open weights and published training documentation.
In our view, Australian AI models should be: • Documented and inspectable, with open weights and published training documentation.
Says model versions used in published research or government decision-making should be accessible or archived in a form permitting verification.
Versioned and archived such that any model version used in published research or in government decision-making is accessible or archived in a form that permits verification.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Support Australian AI capability with documented, inspectable, open-weight models, published training documentation, independent evaluation, and versioning and archiving so versions used in research or government remain accessible or archived for verification.
we strongly support the development of Australian AI capability. In our view, Australian AI models should be: • Documented and inspectable, with open weights and published training documentation. • Versioned and archived such that any model version used in published research or in government decision-making is accessible or archived in a form that permits verification. • Open to independent evaluation such that public trust can be earned, rather than merely gestured at.
Recommend that the research community, including metaresearch and research integrity experts, take a formal and continuing role in AI governance, including setting standards.
we urge the Committee to recommend that the research community, including those with metaresearch and research integrity expertise, take on a formal and continuing role AI governance. This should include both setting standards and ongoing governance.