Human Computer Interaction, Artificial Intelligence and Intelligent Augmentation
Loading...
Paper Type
short
Paper Number
2089
Description
AI-based decision support systems (DSS) have become increasingly popular for solving a variety of tasks in both, low-stake, and high-stake situations. However, due to their complexity, they often lack transparency into their decision process. Therefore, the field of explainable AI (XAI) has emerged to provide explanations for these black-box systems. While XAI research assumes an increase in confidence when using their augmented grey-box systems, test designs for this proposition are scarce. Therefore, we propose an empirical study to test the effect of black-box, grey-box, and white-box explanations on a domain expert’s confidence in the system, and subsequently on the effectiveness of the overall decision process. For this purpose, we derive hypotheses from theory and implement AI-based DSS with XAI augmentations for low-stake and high-stake situations. Further, we provide detailed information on a future survey-based study, which we will conduct to complete this research-in-progress.
Recommended Citation
Wanner, Jonas; Herm, Lukas-Valentin; Heinrich, Kai; Janiesch, Christian; and Zschech, Patrick, "White, Grey, Black: Effects of XAI Augmentation on the Confidence in AI-based Decision Support Systems" (2020). ICIS 2020 Proceedings. 14.
https://aisel.aisnet.org/icis2020/hci_artintel/hci_artintel/14
White, Grey, Black: Effects of XAI Augmentation on the Confidence in AI-based Decision Support Systems
AI-based decision support systems (DSS) have become increasingly popular for solving a variety of tasks in both, low-stake, and high-stake situations. However, due to their complexity, they often lack transparency into their decision process. Therefore, the field of explainable AI (XAI) has emerged to provide explanations for these black-box systems. While XAI research assumes an increase in confidence when using their augmented grey-box systems, test designs for this proposition are scarce. Therefore, we propose an empirical study to test the effect of black-box, grey-box, and white-box explanations on a domain expert’s confidence in the system, and subsequently on the effectiveness of the overall decision process. For this purpose, we derive hypotheses from theory and implement AI-based DSS with XAI augmentations for low-stake and high-stake situations. Further, we provide detailed information on a future survey-based study, which we will conduct to complete this research-in-progress.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.