This paper explores the interplay between algorithmic advice, organizational rules, and discretionary practices in an decision-making context, focusing on loan evaluation processes Specifically, we investigate how algorithmic output, presented as a risk score, is paired with company guidelines that recommend the appropriate screening method—either video calls or face-to-face meetings—to the loan evaluator. By applying paradox theory and examining compliance with those guidelines through the lenses of automation and augmentation, we analyze how loan evaluators’ adherence to these recommendations moderates loan decision outcomes. We utilize a novel dataset of 20,734 loan applications, with a focused subsample of 2,601. Utilizing a novel dataset of 20,734 loan applications, with a focused subsample of 2,601, our findings reveal that evaluator experience inversely correlates with the adoption of prescribed screening practices. This suggests that a nuanced approach to integrating algorithms is essential to leverage their benefits while preserving professional discretion. This study contributes to the use of algorithmic advice in expertise and entrepreneurial finance literature understanding the complex dynamics of algorithmic decision-making in organizational contexts, offering insights into balancing efficiency with equity in automated systems.
