Union wants considered rollout of AI to temper concerns

The Community and Public Service Union has called for a cautious approach to the rollout of artificial intelligence in government agencies, warning that its benefits should be viewed with scrutiny. At a federal inquiry into AI, which has resumed its 11th hearing, union leader Rebecca Fawcett emphasized the potential risks of misinterpreting AI's capacity as budget savings.

The union expressed concern over the uncertain costs associated with AI implementation in businesses, arguing that without reasonable returns on investment, government expenditures could rise unproductively. The submission highlights a UK analysis estimating billions in savings from AI-driven public service delivery, yet it raises questions about the reliability of existing AI models.

Fawcett stressed that efficiency improvements must not compromise the accuracy of decisions or the reliability of public services. She advocated for a locally controlled AI capability to safeguard Australians' data and prevent overseas access, following warnings from the Australian Signals Directorate about the urgent need to fortify government systems against AI-fueled cyber threats.

The union also proposed a data trust model to ensure fair compensation for Australian creators whose data is utilized in training AI, suggesting that such a system could prevent value from being monopolized by a few multinational corporations.