Mapping AI's Role in NSW Governance: A Socio-Technical Analysis of GenAI Integration

Luis Lozano-Paredes^*

• TD School, University of Technology Sydney, Sydney, Australia

This paper examines the integration of Generative Artificial Intelligence (GenAI), particularly large language models (LLMs), in public governance within the New South Wales (NSW) Government of Australia. Using Actor-Network Theory (ANT) as a methodological-theoretical framework alongside concepts of breakdown and repair, the study conceptualises LLMs as non-human actants within government networks that actively reshape relationships, workflows, and decision-making processes. Through document analysis, actor-network mapping, and scenario analysis, the research identifies potential vulnerabilities in GenAI implementation, including transparency challenges, algorithmic bias, and governance conflicts. The study analyses NSW's AI Strategy, Ethics Policy, and workforce training initiatives, with urban planning presented as a practical application domain. The paper concludes that effective AI governance requires both robust technical frameworks and a fundamental reconsideration of how agency and accountability are distributed across human-AI networks, emphasising the importance of maintaining democratic oversight, human judgment, and public contestability as LLMs become increasingly embedded in administrative processes.