Beyond Automation: Algorithm Stewardship as Organizational Capability in AI Adoption - A Multi-Level Analysis Across HR and Procurement in Tangier’s Automotive Cluster

Abstract

This exploratory study examines artificial intelligence (AI) adoption in Human Resources and Procurement functions within Tangier's automotive cluster, addressing tensions between efficiency-oriented and capability-based perspectives. We investigated six automotive firms through twenty semi-structured interviews and document triangulation, integrating Resource-Based View, Technology Acceptance Model, and sociotechnical systems theory. Small sample (n=6) and 50% failure rate limit generalizability; findings represent illustrative configurations rather than patterns. Organizations demonstrating mature "algorithm stewardship" integrating technical competency, governance routines, and ethical reflexivity observed substantial improvements in successful cases: time-to-hire reductions up to 34% (corroborated), turnover prediction accuracies 74% (self-reported), forecasting improvements 16%. However, half experienced implementation failures due to insufficient data, inadequate governance or system misalignment. HR prioritized transparency and bias mitigation; Procurement emphasized accuracy and risk reduction. Cross-functional governance structures co-occurred with success, though causality remains indeterminate. Small sample precludes generalization and causal inference. Performance represents best-case scenarios from well-resourced implementations. Despite limitations, we extend Resource-Based View through algorithm stewardship conceptualization, refine Technology Acceptance Model via functional heterogeneity, and identify sociotechnical alignment mechanisms. Tentative 18-month implementation framework requires validation. Organizations lacking data maturity, governance capabilities, or technical expertise face elevated failure risk. First empirical AI adoption comparison across HR and Procurement in emerging automotive cluster. For accounting scholarship, we demonstrate AI necessitates extensions to management accounting systems, internal controls, and accountability mechanisms.