Carolin Leluschko’s research while affiliated with Deutsches Forschungszentrum für Künstliche Intelligenz and other places

A currently upcoming direction in the research of explainable artificial intelligence (XAI) is focusing on the involvement of stakeholders to achieve human-centered explanations. This work conducts a structured literature review to asses the current state of stakeholder involvement when applying XAI methods to remotely sensed image data. Additionally it is assessed, which goals are pursued for integrating explainability. The results show that there is no intentional stakeholder involvement. The majority of work is focused on improving the models performance and gaining insights into the models internal properties, which mostly benefits developers. Closing, future research directions, that emerged from the results of this work, are highlighted.