Artificial intelligence (AI)-based information systems are increasingly integrated into operational decision-making, particularly in data-intensive industrial organizations where decisions are frequent, time-sensitive, and closely connected to everyday work practices. While previous research has emphasized AI’s potential to improve efficiency, accuracy, and decision support, less attention has been paid to the tensions that emerge when AI systems interact with human decision-making practices in operational contexts. This study therefore explores what tensions emerge in the use of AI-based information systems for operational decision-making. The study is based on a qualitative interview study with a sample of 12 participants from four organizations operating in data-intensive industrial environments. Empirical data were collected through semi-structured interviews with practitioners who use, manage, develop, or advise on AI-based systems. The data were analyzed through an abductive approach supported with grounded theory principles, combining empirical coding with theoretical concepts related to human-AI collaboration, operational decision-making, and organizational tensions. The findings show that AI-based information systems are mainly used to support rather than replace human decision-makers. AI contributes to operational decision-making by improving information processing, automating repetitive tasks, supporting analysis, and generating recommendations. However, the integration of AI also creates tensions that must be managed in practice. Three central tensions were identified: control versus flexibility, automation versus professional judgment, and efficiency versus responsibility. These tensions demonstrate that AI-supported operational decision-making is not a straightforward process of technological optimization. Instead, it requires continuous balancing between the analytical capabilities of AI systems and the contextual understanding, judgment, and accountability of human actors. The study concludes that AI-based information systems reshape rather than eliminate human decision-making in operational work. Human involvement shifts from manual execution toward supervision, interpretation, validation, and responsibility for AI-generated outputs. The study contributes to information systems research by providing a context-sensitive understanding of how AI-based information systems influence operational decision-making and how practitioners navigate the tensions that arise in everyday organizational practice.