DHCRC localisation of OECD AI Classification Framework to support governance of investment and commercialisation of AI systems amongst Australian SMEs

Start Date

13 November 2024

End Date

30 September 2025

Project Participants

Australian Government Department of Health Disability and Ageing, University of Technology Sydney, UTS Human Technology Institute, UTS Rapido

Status: Completed

Tags: AI governance, artificial intelligence (AI), classification systems, health system, innovation

Opportunity 

AI has the potential to improve productivity, decision-making and service delivery across healthcare. However, responsible adoption depends on developers, health organisations and policymakers having a shared understanding of how AI systems are designed, deployed and governed. The Organisation for Economic Cooperation and Development (OECD) AI Classification Framework offers an international foundation for describing Artificial Intelligence (AI) systems, but its original static and domain-agnostic format did not reflect Australian standards, laws or the specific needs of healthcare.  

This project explored how to enable the implementation of the OECD AI Classification Framework in support of responsible stewardship of trustworthy AI in Australian healthcare by localising the framework and making it accessible through an online tool.  

Objectives 

This project aimed to develop an easy-to-use and intuitive web-based tool to help classify different types of AI solutions used in Australian healthcare by adapting the OECD AI Classification Framework to the Australian context, reflecting recent government policies including proposed mandatory guardrails for AI across the Australian economy. The tool would complement efforts by governments, peak organisations, and clinical professional and safety bodies to ensure AI is deployed safely into healthcare settings.  

The collaboration delivered a functional prototype web tool that applied the OECD AI Classification Framework to the Australian healthcare sector through a health-specific question and answer set, referring to authoritative Australian resources for contextual support for end-users, and providing a dynamically-generated classification report as a result. The prototype was tested by end users and sector experts who provided feedback on the tool through focus groups and expert advisory sessions respectively. 

The business logic for identifying potential governance focus areas, the development of the prototype, and the findings from the testing of the tool was documented in a detailed research report and roadmap for development and implementation. 

The project is a starting point for providing Australian health organisations with a practical approach to describing AI systems consistently for sector alignment when considering key governance issues such as human impact, privacy, transparency, meaningful choice, human oversight, system performance and workforce readiness. By connecting the OECD AI Classification Framework with Australian standards, policies and health-sector guidance, the project demonstrated how responsible AI principles can be translated into an accessible digital tool for healthcare stakeholders locally. 

A follow-on project Localisation of the OECD AI Classification Framework (AICF) for the Australian Health Sector – Phase 2 is continuing the work of enhancing a prototype tool and validating it using real-world healthcare case studies to strengthen its ability to support consistent AI governance, procurement, validation and evidence-based adoption across the Australian health sector.  

Together, these projects represent a progressive pathway toward a more robust and trusted approach to AI governance in healthcare, supporting organisations to better understand, classify and assess AI systems as part of responsible AI adoption.

Integrity, Excellence,
Teamwork and Authenticity

Hand holding a smartphone against a colourful, defocused background