Downloads & Free Reading Options - Results

The role of trust is an essential area of research for explainable AI, where trust is measured by the human's willingness to use the machine's suggestions. There are several existing studies that investigate and show that human-AI team performance on decision-making tasks improves when the AI explains its recommendations. However, most of these studies focus on binary decision-making tasks where the human and the machine have near-equal performance outcomes. This is different from a typical visualization scenario as analysts might want to explore the data and gather their own insights. In this project, we conduct user studies to understand the factors that influence trust in a data exploration setting where the task complements both the strengths of the human and the machine. As visual interfaces for data exploration already display a lot of information, we investigate the minimal amount of explanations needed for building trust and encouraging suggestion usage while minimizing information overload. We also seek to understand whether there truly is a correlation between the trust perceptions of the user and the usage of AI/ML suggestions. Explanations of the suggestions will be shown to users through a combination of keywords and confidence percentages.