How public servants use Generative AI in their work
How public servants use Generative AI in their work
Generative AI (GenAI) promises to deliver positive impacts on productivity and governments are under pressure to use these tools, despite the risks they raise. An article in the Australian Journal of Public Administration examines how Australian public servants understand and use GenAI in their work. It argues that GenAI adoption involves relational processes that require moving beyond treating technology as a fixed entity to examining the evolving relations between users, technologies and organisational contexts.
A relational process
The adoption of GenAI has been swift across sectors, driven largely by promises of efficiency and productivity gains and the potential it holds in driving innovation in fields like business, medical research, health care, engineering, and technology. Public sectors globally have also begun experimenting with a range of different uses of GenAI.
GenAI adoption fundamentally involves relational processes that require moving beyond treating technology as a fixed entity to examine the evolving relations between users, technologies, and organisational contexts. This necessitates understanding how users make sense of this disruptive technology within their specific contexts and the changes that may occur in workplace practices (e.g. human–AI collaboration, teams, roles, and interpersonal dynamics).
About the research
To explore GenAI adoption in the Australian public sector, the researchers conducted 37 semi-structured interviews across federal, state, and territory jurisdictions from November 2024 to January 2025. Given the nascent nature of GenAI, interviews were selected as the primary method to capture the lived experiences and perceptions of public servants navigating this technological disruption.
The tensions and complexity of GenAI
The research shows that while GenAI uptake is gathering momentum across governments, implementation approaches remain highly differentiated. The findings reveal inherent tensions that highlight the complexity GenAI introduces to public sector adoption.
1. When technological momentum meets knowledge ambiguity
There was a sense of inevitability that GenAI technologies and their integration into organisational life were going to be the new norm. In some instances, this inevitability guided more proactive approaches to GenAI adoption with agencies deliberately trying to foster experimentation.
Despite this momentum, there are concerns regarding knowledge ambiguity around GenAI use in the public sector. It reveals a striking paradox where organisations feel compelled to adopt technologies, they do not fully understand or whose impacts they cannot adequately predict.
2. Managing the tension between agile regulation and bureaucratic prudence
Issues of privacy, data security and algorithmic accountability are bound with GenAI, especially as transparency, accountability and public trust are paramount to public service. Consequently, regulation and regulatory oversight were significant concerns for some public servants.
Participants identified risks such as the quality of output and bias and the breakdown of administrative accountability. Acknowledging both the need to adopt GenAI and the appropriate governance frameworks was seen as a pragmatic response. However, respondents were aware of the slow-moving nature of government. The slow regulatory response is often characterised by processes such as consensus building across different stakeholders and by risk aversion.
3. Using GenAI within and around administrative constraints
Public servants made everyday use of GenAI including writing emails, developing first drafts of policy briefs, creating econometric models, and analysing submissions in consultation processes. Efficiency was the rationale underlying these uses.
In some cases, there were officially endorsed pathways to explore and experiment but there was also unsanctioned use. The latter demonstrates the willingness of public servants to circumvent restrictions while calling for cautious implementation. It suggests that GenAI technology operates not only to be regulated, but as a relational partner that changes the knowledge work of public servants.
4. Preserving legitimacy while adopting GenAI
As public servants gradually incorporate GenAI into their daily practice, they drew attention to the need to ensure tools are used critically while maintaining legitimacy. Empirical data showed public servants shared a collective concern about maintaining citizen and community trust as GenAI is adopted. There was a recognition that preserving trust and legitimacy requires responsible use of GenAI technologies. There was also a concern that GenAI is being used uncritically, and AI outputs could be accepted without question.
The bottom line
GenAI implementation in government is complex. GenAI adoption should be viewed through a relational process lens with interconnected social and technical factors - GenAI platforms, the public servants who use them, entrenched bureaucratic values and priorities, existing risk frameworks, and governing laws and policies. All are parts of an intricate system. Public servants recognise that critical, human-centred use of GenAI can enhance and help maintain legitimacy, but the technology's design and presentation can encourage uncritical adoption that threatens these same values.
Want to read more?
Navigating complexity: A relational perspective on generative AI adoption in government - Shibaab Rahman, James Connor, Helen Dickinson, Kate Henne and Vanessa McDermott, Australian Journal of Public Administration, April 2026
Read another Research Brief on this topic