SkandAI: Generativ artificiellintelligens för kapitalförvaltning : En socioteknisk studie av möjligheter, begränsningaroch ansvar

Sammanfattning

This study addresses the limited empirical research on GenAI in financial analysis bydeveloping a retrieval augmented generation (RAG) pipeline to examine its practical potentialwithin capital management. To situate the technical artefact in a broader context, the studyextends the Design Science Research (DSR) framework into a Meta-DSR approach, integratingtechnical development with a qualitative contextual investigation of the Swedish capitalmanagement market. Interviews conducted with actors across the financial market reveal abroadly positive outlook toward GenAI and its potential use cases, while highlightingaccountability as a central concern when GenAI is introduced into financial processes. Placed inits broader contextual environment, the locally deployed RAG pipeline offers greater traceabilityand control compared to third-party alternatives. These are qualities that carry strategicsignificance in an industry where competitive advantage is tied to the handling of sensitiveinformation. This broadens the discussion beyond technical performance as the sole metric ofevaluation and raises the question of whether these structural benefits may outweigh strictlytechnical performance for financial organisations. The study concludes that the value of GenAIin capital management is not general but context-dependent, and that successfulimplementation requires equal attention to organisational, strategic, and accountabilityconsiderations as to technical innovation. Regarding the design of a GenAI-based tool forfinancial report analysis, the findings suggest that transparency and controllability are equalparameters to raw technical performance as key design parameters. More broadly, theopportunities and barriers shaping GenAI adoption in capital management are found to be lesstechnical than organisational, rooted in questions of accountability, knowledge, and strategicdirection. Directions for future research include examining full-scale implementation and thelong-term organisational consequences of integrating GenAI into financial workflows.