Overcoming Algorithm Aversion with Transparency: Can Transparent Predictions Change User Behavior?

TL;DR

This study investigates whether transparency in machine learning models, combined with user adjustability of predictions, can reduce algorithm aversion, finding that adjustability helps but transparency's role is limited and largely independent.

Contribution

It replicates prior findings on adjustability reducing algorithm aversion and extends them by examining the role of interpretability and transparency in this context.

Findings

Adjustability mitigates algorithm aversion.

Transparency had a smaller, non-significant effect.

Transparency and adjustability effects are largely independent.

Abstract

Previous work has shown that allowing users to adjust a machine learning (ML) model's predictions can reduce aversion to imperfect algorithmic decisions. However, these results were obtained in situations where users had no information about the model's reasoning. Thus, it remains unclear whether interpretable ML models could further reduce algorithm aversion or even render adjustability obsolete. In this paper, we conceptually replicate a well-known study that examines the effect of adjustable predictions on algorithm aversion and extend it by introducing an interpretable ML model that visually reveals its decision logic. Through a pre-registered user study with 280 participants, we investigate how transparency interacts with adjustability in reducing aversion to algorithmic decision-making. Our results replicate the adjustability effect, showing that allowing users to modify algorithmic predictions mitigates aversion. Transparency's impact appears smaller than expected and was not significant for our sample. Furthermore, the effects of transparency and adjustability appear to be more independent than expected.