Exploring the Data Sourcing Decision for AI – A Delphi Study

Access to abundant, high-quality training data is paramount for successfully implementing artificial intelligence (AI) in organizations. However, internal data resources are often insufficient, leading organizations to seek external data sources to enhance AI system performance, robustness, and value contribution. Despite extensive research on outsourcing decisions in the information systems (IS) literature, the focus on data as a primary outsourcing object has remained limited. This gap is significant given the data’s unique ontological and economic characteristics as an organizational asset, which challenge traditional economic assumptions in IS sourcing theory. This Delphi study identifies and discusses nine context-specific determinants that experts in our panel consider most important for sourcing external AI training data. The discussion of our findings builds on and expands the perspectives of Transaction Cost Economics (TCE) and the Resource-Based View (RBV), contributing to the intersection of the emerging IS sourcing literature sub-streams on AI and data sourcing.