Reframing Financial Decision-Making in the Age of AI : Rethinking Oversight, Accountability, and Control in the Algorithmic Era

Sammanfattning

Title: Reframing Financial Decision-Making in the Age of AISubtitle: Rethinking Oversight, Accountability, and Control in the Algorithmic Era Financial decision-making (FDM) workflows are being reconfigured in light of the proliferation of artificial intelligence (AI) technologies. This thesis investigates how the transformations witnessed across sectors and industries are reshaping oversight, accountability, and control structures as AI gains influence on budgeting, forecasting, and financial planning. Assumptions about human agency and governance are fundamentally challenged as organizations move towards an AI-augmented environment across its verticals. This research objective is twofold: first, to critically examine the relevance and applicability of financial decision making theories such as Accounting Information & Managerial Work (Hall, 2010) and In Search of a Theory of Budgeting (Kenno et al., 2018) to the AI-enabled financial landscape; and second, to assess how a hybrid decision-making environment through Human-AI Ensembles (Choudhary et al., 2023) and Accountability in Algorithmic Decision Making (Diakopoulos, 2016) is reconceptualizing longstanding decision structures. Through a qualitative analysis of the semi-structured interviews conducted with management and technology implementation consultants, this thesis identifies the major shifts organizations are experiencing on various fronts, most notably the emergence of hybrid governance structures and unraveling of traditional financial logic in algorithmic contexts. The findings derived from both primary and secondary data reveal three core phenomena:1) Epistemic decoupling: Managers find themselves relying on, overseeing, and interpreting model outputs that they cannot comprehensively validate or audit.2) Networked accountability: Responsibility is continuously negotiated across human and technological actors.3) Polycentric control: Where decision-making authority and control become fragmented across human and technological agents. In sum, this thesis contributes to the academic discussion on AI-augmented organizational landscapes by proposing a theoretical reconceptualization of financial decision-making under the influence of AI and algorithmic actors. Additionally, institutional adaptations are outlined throughout this study that are required in order to sustain the validity, legitimacy, and transparency of financial governance in the algorithmic era.