Achieving trustworthy AI in the public sector: A maturity index
Acceptance of AI technologies relies on building and sustaining stakeholder trust. Given the risks associated with AI, Trustworthy AI has become a global topic of discussion, and the European Commission’s influential principles for trustworthy AI recently informed Australia’s AI Ethics Framework.
However, despite broad consensus on these principles, a gap remains between these principles and how AI systems are developed and used in practice. In response, governments and industry have urgently called for research on how AI can be deployed in a trustworthy way.
This project focuses on public sector use of AI to:
- Develop an empirical understanding of how organisations can proactively achieve trustworthy AI-enabled public services and mitigate risks to stakeholders; and
- Develop and validate a tool to assess organisational maturity in achieving trustworthy AI
This tool will enable public sector and government agencies to assess the maturity of their processes and practices to produce trustworthy AI and adhere to Australia’s ethical AI principles.
The development of this tool will be informed by UQ’s recent research reports on trustworthy AI and public trust in AI (in partnership with KPMG), coupled with a rigorous empirical field examination of AI systems currently in use in public sector service delivery.
After data collection and analysis of the field data, the survey will be developed and piloted to ensure robust psychometric properties and then validated by examining how the tool predicts stakeholder trust, acceptance and willingness to use the AI systems.
The survey will be developed into a self-assessment tool with supporting guidebook to enable public sector leaders and managers to:
a) self-diagnose gaps between their practices and the principles of trustworthy and ethical AI
b) understand best-practice approaches to close the gaps; and
c) benchmark and monitor stakeholder trust in the organisation’s AI-enabled public services over time to provide regular feedback.
A multistakeholder roundtable will be held to facilitate translation into policy and practice.