392246 Projekt: Human-Centered Evaluation of XAI-Assisted Decision Making (Pj) (WiSe 2024/2025)

AI is increasingly being used in various domains such as healthcare, finance, and recruitment to augment complex decision-making processes. Research in AI-assisted decision-making aims to improve practitioners' decision-making capabilities by including additional information from data-driven AI approaches. Many of these tools incorporate explainable AI to help improve user understanding of model decisions. However, the effectiveness of different XAI approaches for this goal is not well understood.
The goal of this project is to create an online interface designed to evaluate various XAI approaches within the context of AI-assisted decision making. The tool will function as an interactive AI-based decision support system, that allows for the configuration of varying XAI methods based on specific data modalities and research questions. Additionally, to validate the efficacy of the tool in comparing different XAI methodologies, a small user study will be conducted as part of the project.
The project has multiple components where students can contribute and gain firsthand knowledge. In addition to web development, students should have some knowledge of machine learning, explainable AI, and preferably basic knowledge of human-centered approaches.
This project is suitable for 3 students due to its complexity but can be scaled down for 1 or 2 students as needed.