This article investigates how a data-driven public sector organisation attempts to implement artificial intelligence (AI) and generative AI (GenAI). Public sector organisations face considerable pressure to adopt AI, which is driven by promises about its transformative potential. However, considerable uncertainty remains about how these organisations can begin developing AI solutions to realise the supposed benefits.
By employing an analytical autoethnographic approach (Anderson, 2006), this study positions the author as a member of the organisation under study and a qualitative researcher. Considering this dual role provides a unique perspective for understanding an organisation's engagement with AI, especially in terms of its internal development processes.
The study shows that an organisation initiated a pilot project to explore AI. The project responds to a perceived need to “take a stance on AI” by formulating several relevant cases of AI use. It involves collaboration with an external tech provider to develop AI prototypes. In the end, as no AI solutions are derived, the pilot project turns out to be somewhat of a failure. Concurrently, however, it provides the organisation with much-needed insights into and learning about the potentials and pitfalls of AI.
The study emphasises Suchman's concept of questioning the “thingness” of AI, challenging the perception of AI as a stable, agent-like entity (Suchman, 2023). The originality of the study lies in how it applies Suchman's ideas to a real-world case of AI implementation, showing how the organisation persistently tried to give AI a presence within its operations while grappling with the difficulty of achieving any tangible “thingness” related to AI.
