The Risk of Algorithm Transparency: How Algorithm Complexity Drives the Effects on Use of Advice

Algorithmic decision support is omnipresent in many managerial tasks, but human judgment often makes the final call. A lack of algorithm transparency is often stated as a barrier to successful human-machine collaboration. In this paper, we analyze the effects of algorithm transparency on the use of advice from algorithms with different degrees of complexity. We conduct a preregistered laboratory experiment where participants receive identical advice from algorithms with different levels of transparency and complexity. The results of the experiment show that increasing the transparency of a simple algorithm reduces the use of advice, while increasing the transparency of a complex algorithm increases it. Our results also indicate that the individually perceived appropriateness of algorithmic complexity moderates the effects of transparency on the use of advice. While perceiving an algorithm as too simple severely harms the use of its advice, the perception of an algorithm being too complex has no significant effect on it. Our results suggest that managers do not have to be concerned about revealing complex algorithms to decision makers, even if the decision makers do not fully comprehend them. However, making simple algorithms transparent bears the risk of disappointing people’s expectations, which can reduce the use of algorithms' advice.