AI Inquiry Atlas

An unofficial, independent guide to the public records of the Joint Select Committee on Artificial Intelligence. Not connected with the Parliament of Australia or the committee. Its makers also made a submission (no. 481; see About). Summaries and scores are written by AI and can be wrong: check the original before relying on them.

Submission 53

ADEIA, UNSW and the ARC Centre for Next-Gen Architectural Manufacturing

Submitter group: Industry bodies.

An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.

Open in the interactive guide

Summary

Read the original submission for the full text.

Argues the built environment should be a core part of national AI policy, with AI augmenting rather than replacing accountable professionals. Addresses SME adoption, sovereign domain models and data, workforce pathways, Commonwealth procurement, copyright and Indigenous data, and a built-environment stream in the AI Safety Institute.

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.

  1. Argues that responsibility for consequential decisions cannot be delegated to AI and requires a competent professional with authority to exercise independent judgement.

    For consequential decisions, responsibility cannot be delegated to an AI system. The accountable professional must have the competence, authority and information necessary to exercise independent judgement and must be able to explain and defend the decision made.

  2. Argues that the built environment should be treated as a horizontal enabling sector within national AI policy.

    The built environment should be treated as a horizontal enabling sector within national AI policy

  3. Describes barriers for built-environment SMEs including system costs, limited time for legal and cyber assessment, and fragmented data.

    Common barriers for built-environment SMEs, including owner-led businesses, include the cost of enterprise systems, limited time for legal and cyber assessment, fragmented project data, skills shortages, uncertainty about professional indemnity insurance

  4. Argues that sovereign capability is more than server location and includes control over critical data, portability and a viable provider exit.

    Sovereign capability should not be reduced to server location. It includes Australian control over critical data and expertise, lawful access, technical and commercial portability, provenance, model and version records, local research capability, competition and a viable exit from a provider.

  5. Argues that practices automating foundational tasks should identify lost experience and provide an equivalent supervised learning pathway.

    Practices that automate foundational tasks should identify the experience being removed and provide an equivalent pathway through supervised manual checking, rotations, site exposure, detailing, code research, design critique, client communication and peer review.

  6. Argues that the Commonwealth should act as a model client through procurement terms covering disclosure, data rights and audit access.

    Procurement should make the Commonwealth a model client. Tender and contract documents should define permitted uses, material disclosure, project-data and training rights, confidentiality, ICIP, cyber security, model and version records, independent review, incident notification, audit access, provider changes, data return, portability and exit.

Recommendations

Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.

  1. Recognise the built environment as a sector of comparative advantage and establish a national Built Environment AI Mission, with a standing cross-jurisdictional forum and demonstrators measuring safety, quality, sustainability, regional benefit and commercialisation.

    Recognise the built environment as a sector of comparative advantage and establish a national Built Environment AI Mission, supported by a standing cross-jurisdictional forum and demonstrators that measure safety, design quality, sustainability, regional benefit and commercialisation.

  2. Support narrow, domain-specific AI systems and governed data environments for Australian building codes and standards, planning controls, climate and hazard data, materials, procurement, asset performance and infrastructure management.

    Australia should support narrow, domain-specific AI systems and governed data environments for Australian building codes and standards, planning controls, climate and hazard data, materials and product information, procurement requirements, asset performance and infrastructure management.

  3. Make supported domain systems professional decision-support rather than automated certification, retaining identifiable current authoritative sources and human professional responsibility.

    These systems should be designed as professional decision-support, not automated certification. Authoritative sources must remain identifiable and current; human professionals must retain responsibility;

  4. Have supported domain-system data access respect copyright, confidentiality, privacy, security and Indigenous Cultural and Intellectual Property.

    and data access must respect copyright, confidentiality, privacy, security, Indigenous Cultural and Intellectual Property (ICIP)

  5. Have publicly funded datasets and models use interoperable formats and clear licensing so Australian SMEs and researchers can reuse them rather than one platform capturing them.

    Publicly funded datasets and models should use interoperable formats and clear licensing so that capability can be reused by Australian SMEs and researchers rather than captured by one platform.

  6. Commission an annual built-environment AI adoption and incident baseline.

    Commission an annual built-environment AI adoption and incident baseline

  7. Fund shared capability hubs for small and regional practices.

    fund shared capability hubs for small and regional practices

  8. Establish governed Australian domain-model and data-environment pilots with provenance, interoperability, rights, security and professional-assurance requirements.

    establish governed Australian domain-model and data-environment pilots with provenance, interoperability, rights, security and professional-assurance requirements.

  9. Extend education/workforce policy beyond prompts: probabilistic failure, primary-source checks, data/privacy/cyber, copyright/ICIP, bias/accessibility, records/disclosure/incidents, ethics, critique, communication and safe non-AI work; demonstrate competence through role-relevant scenarios and supervised practice.

    Education and workforce policy should extend beyond prompt use. Required capability includes understanding probabilistic failure; checking primary sources; data stewardship, privacy and cyber security; copyright and ICIP; bias and accessibility; records, disclosure and incident reporting; ethical reasoning; design critique; communication; and the ability to work safely without AI. Competence should be demonstrated through role-relevant scenarios and supervised practice, not attendance alone.

  10. Establish a built-environment AI compact across universities, VET, registration/accreditation bodies, employers and professional organisations, with funded SME access, educator-practitioner exchange, role-based competence and explicit graduate skills-preservation plans.

    Establish a built-environment AI capability compact across universities, VET, registration and accreditation bodies, employers and professional organisations, with funded SME access, educator-practitioner exchange, role-based competence and explicit graduate skills-preservation plans.

  11. Commonwealth entities commissioning buildings, infrastructure, planning or asset services should identify a named accountable official and a named qualified professional for each material AI use.

    Commonwealth entities commissioning buildings, infrastructure, planning or asset services should identify a named accountable official and a named qualified professional for each material AI use.

  12. Make Commonwealth material-AI transparency describe use/purpose, affected decision, system/service category, human control point, data/rights issues, assurance and review/complaint pathway.

    Public transparency should be meaningful rather than generic: it should describe the use and purpose, the decision affected, the system or service category, the human control point, the principal data and rights issues, the assurance undertaken and the pathway for review or complaint.

  13. Set Commonwealth tender/contract terms for permitted uses, material disclosure, project-data/training rights, confidentiality, ICIP, cyber security, model/version records, independent review, incidents, audits, provider changes, data return, portability and exit.

    Procurement should make the Commonwealth a model client. Tender and contract documents should define permitted uses, material disclosure, project-data and training rights, confidentiality, ICIP, cyber security, model and version records, independent review, incident notification, audit access, provider changes, data return, portability and exit.

  14. Reward whole-of-life value and Australian capability in procurement evaluation; avoid transferring unlimited or uninsurable provider-controlled risk to small professional firms.

    Evaluation should reward whole-of-life value and Australian capability, and should not transfer unlimited or uninsurable provider-controlled risk to small professional firms.

  15. Extend the Government AI policy into a model built-environment procurement and assurance schedule.

    Extend the Government AI policy into a model built-environment procurement and assurance schedule

    A program corrected obvious text-recognition errors in this quote for display.

  16. Require public registers and impact assessments for material AI uses in Commonwealth planning, design, construction, property and infrastructure decisions.

    require public registers and impact assessments for material AI uses in Commonwealth planning, design, construction, property and infrastructure decisions.

    A program corrected obvious text-recognition errors in this quote for display.

  17. Require high-risk built-environment uses, including statutory submissions, fire, structure, access, code compliance, certification and consequential agentic actions, to have pre-approval, independent checks, traceability, revalidation after material changes and incident/withdrawal/correction processes.

    High-risk uses - including statutory submissions, fire, structure, access, code compliance, certification and consequential agentic actions - should require documented pre-approval, independent checking, traceability, revalidation after material changes and an incident, withdrawal and correction process.

  18. Examine rights of contribution and targeted evidentiary consequences where a provider controls information needed to investigate a failure.

    Government should examine rights of contribution and targeted evidentiary consequences where a provider controls information needed to investigate a failure.

  19. Have the Attorney-General's Copyright and AI processes expressly include drawings, specifications, reports, images, digital models, project datasets and digital twins.

    The Attorney- General's Copyright and AI Reference Group is considering lawful licensing, certainty for AI-generated material and lower-cost enforcement. Those processes should expressly include drawings, specifications, reports, images, digital models, project datasets and digital twins.

    A program corrected obvious text-recognition errors in this quote for display.

  20. Do not make AI an author; clarify AI-assisted human authorship and support provenance recording human design intent, direction, selection, rejection, revision and final judgement.

    Australian policy should not make AI itself an author. It should clarify Al- assisted human authorship and support a provenance standard recording human design intent, direction, selection, rejection, revision and final judgement.

    A program corrected obvious text-recognition errors in this quote for display.

  21. Apply the NIAA Framework for Governance of Indigenous Data and place First Nations partners in decision-making roles throughout the data lifecycle.

    The Commonwealth should apply the NIAA Framework for Governance of Indigenous Data and place First Nations partners in decision-making roles throughout the data lifecycle.

  22. Clearly label and make traceable material synthetic content used in planning, tenders, sales, consultation or public communication.

    Material synthetic content used in planning, tenders, sales, consultation or public communication should be clearly labelled and traceable.

  23. Provide high-impact built-environment systems with human review, complaint, correction and remedy pathways and monitoring across projects and over time.

    High-impact systems require human review, complaint, correction and remedy pathways, together with monitoring across projects and over time.

  24. Apply heightened safeguards where AI informs schools, hospitals, aged care, accessible housing or other places and services used by children and vulnerable people, including accessibility, safeguarding, consent, contestability and independent professional review.

    Within its field, heightened safeguards are warranted where AI informs schools, hospitals, aged care, accessible housing or other places and services used by children and vulnerable people, including accessibility, safeguarding, consent, contestability and independent professional review.

    A program corrected obvious text-recognition errors in this quote for display.

  25. Create a nationally consistent high-risk built-environment assurance protocol.

    Create a nationally consistent high-risk built-environment assurance protocol

  26. Establish provenance and provider-record duties for built-environment work and embed ICIP and Indigenous Data Sovereignty.

    Create a nationally consistent high-risk built-environment assurance protocol; include ADE works and data in copyright reform; establish provenance and provider-record duties; embed ICIP and Indigenous Data Sovereignty; and require labelling, contestability and remedy for consequential AI-supported decisions and synthetic professional content.

    A program corrected obvious text-recognition errors in this quote for display.

  27. Require labelling, contestability and remedy for consequential AI-supported decisions and synthetic professional content.

    and require labelling, contestability and remedy for consequential AI-supported decisions and synthetic professional content.

    A program corrected obvious text-recognition errors in this quote for display.

  28. Have the AI Safety Institute develop formal sector partnerships with ASD, infrastructure agencies, regulators, universities and professional bodies.

    Australia's AI Safety Institute should develop formal sector partnerships with ASD, infrastructure agencies, regulators, universities and professional bodies.

    A program corrected obvious text-recognition errors in this quote for display.

  29. Have AI Safety Institute testing cover advanced/agentic design, asset-management and critical-infrastructure systems: leakage, prompt/data manipulation, unsafe tools, updates, audit-evidence loss and failure in degraded or disconnected conditions.

    Australia's AI Safety Institute should develop formal sector partnerships with ASD, infrastructure agencies, regulators, universities and professional bodies. Testing should cover advanced and agentic systems used in design, asset management and critical infrastructure, including data leakage, prompt or data manipulation, unsafe tool use, model updates, loss of audit evidence and failure under degraded or disconnected conditions.

    A program corrected obvious text-recognition errors in this quote for display.

  30. Translate AI Safety Institute testing findings into procurement and professional assurance guidance, with information-sharing pathways to regulators and the intelligence and security community where critical-infrastructure or foreign-interference risks arise.

    Testing should cover advanced and agentic systems used in design, asset management and critical infrastructure, including data leakage, prompt or data manipulation, unsafe tool use, model updates, loss of audit evidence and failure under degraded or disconnected conditions. Findings should be translated into procurement and professional assurance guidance, with information-sharing pathways to relevant regulators and the intelligence and security community where critical-infrastructure or foreign-interference risks arise.

  31. Establish a built-environment evaluation stream within the AI Safety Institute, aligned with ASD and critical-infrastructure policy.

    Establish a built-environment evaluation stream within the AI Safety Institute; align it with ASD and critical- infrastructure policy

  32. Require security-by-design, evidence preservation, continuity, portability and exit controls for high-risk professional and public-sector AI systems.

    and require security-by-design, evidence preservation, continuity, portability and exit controls for high-risk professional and public-sector AI systems.

    A program corrected obvious text-recognition errors in this quote for display.

  33. Within 0–12 months, issue interim high-risk assurance and model procurement principles alongside the sector forum/mission, adoption/barriers baseline, SME hubs and workforce compact.

    Recognise the sector and establish the Built Environment AI Forum and mission; commission an ADE adoption and 0-12 months barriers baseline; issue interim high-risk assurance and model procurement principles; commence SME hubs and the workforce compact.

  34. Within 12–24 months, pilot governed domain models/data, integrate accreditation and CPD, introduce provider-record/incident duties for professional systems, and test procurement, provenance and impact-assessment templates on public projects.

    Pilot governed Australian domain models and data environments; integrate accreditation and CPD; introduce 12-24 months provider record and incident duties for professional systems; test procurement, provenance and impact-assessment templates on public projects.

  35. Monitor adoption, productivity, job quality, unequal outcomes, cyber incidents, provider concentration and public value; publish de-identified lessons and update standards as models, law and practice change.

    Monitor adoption, productivity, job quality, unequal outcomes, cyber incidents, provider concentration and public Ongoing value; publish de-identified lessons and update standards as models, law and professional practice change.

Read the original

In some quotes a program corrected obvious text-recognition errors for display.

Open in the interactive guide