Analysis · Technology
An AI System Can Go Live and Still Miss Its Goals
By AWEI · AI-compiled · Published · Analysis prepared · 1 source · doi.org
Two Swiss public organizations had AI chatbots in operation, yet reported that their projects had not achieved their goals. That detail in a 2022 study by Oliver Neumann, Katharina Guirguis and Reto Steiner exposes a useful distinction: putting AI into service and demonstrating its value are separate milestones.
The researchers compared eight Swiss public organizations, including ministries, agencies, municipal administrations and state-owned enterprises. They conducted 17 interviews between August 2020 and July 2021 with people involved in AI projects. Interview transcripts were coded for technological, organizational and environmental factors, while allowing additional themes to emerge.
Their central methodological choice was to compare organizations at different levels of AI maturity. An organization exploring its first application was assessed separately from one with established processes for managing many projects. This revealed patterns that a single checklist of adoption barriers could obscure.
In the least experienced organizations, motivated staff, project structure and external partners stood out. At intermediate maturity, management support, budgets, employees and strategic alignment became more prominent. In the most experienced organizations, friction between the units developing AI and those expected to use it emerged as another concern. Collaboration remained important across all three groups.
These were qualitative patterns and proposed explanations, not proof that particular management practices caused success. The researchers compared different organizations rather than following the same organizations through every stage. All cases were Swiss and already working on AI; findings reflected project participants’ assessments. Organization size, legal form and application type also overlapped with maturity, making their separate contributions difficult to establish.
For a public manager, the study offers a way to sharpen a project review: assess deployment status, goal attainment and organizational readiness separately. Then examine the support needed for the next stage. A working chatbot demonstrates implementation; evidence that it meets its intended goals requires a further check.
Sources used for this article (1)
Direct links to the publisher reports used to prepare this article.
- Source 1
- Exploring artificial intelligence adoption in public organizations: a comparative case study Oliver Neumann, Katharina Guirguis, Reto Steiner
