This study aims to examine how different artificial intelligence (AI)-driven knowledge configurations in public organizations shape public value outcomes. Drawing on organizational information processing theory (OIPT), it investigates whether the relational orientation of information flows, process typology and AI technical classification differentially segment three public value outcomes: improved administrative efficiency, improved public service delivery and open government.
The study analyses 418 AI application cases drawn from the European Commission Joint Research Centre’s Public Sector Tech Watch data set. A chi-square automatic interaction detection decision-tree analysis is used to identify the configurational dimensions most strongly associated with public value outcomes.
The results show that the relational orientation of information flows, operationalized through the interaction dimension, is the dominant observed configurational splitter of public value outcomes within this taxonomy-based data set. AI applications oriented toward intragovernmental interaction are overwhelmingly associated with administrative efficiency, whereas those oriented toward external actors are predominantly associated with public service improvement. Process typology and AI technical classification are significantly associated with outcomes at the bivariate level but do not add independent explanatory power in the multivariate configurational model. Open government could not be adequately assessed because of the rarity of positive cases.
The study extends OIPT to AI-enabled knowledge management in the public sector and offers a theory-informed examination of the configurational assumptions embedded in the joint research centre taxonomy. It contributes by showing that, within this taxonomic framework, interaction orientation is the key observed dimension through which AI knowledge architectures are associated with public value creation.
