Explainable artificial intelligence and the social sciences: a plea for interdisciplinary research

Recent research emphasizes the complexity of providing useful explanations of computer-generated output. In developing an explanation-generating tool, the computer scientist should take a user-centered perspective, while taking into account the user’s susceptibility to certain biases. The purpose of this paper is to expand the research results on explainability from the social sciences, and to indicate how these results are relevant to the field of XAI. This is done through the presentation of two surveys to university students. The analysis of the results leads to some interesting hypotheses, for example that the presented order of historical facts might be more influential on the interpretation or appreciation of an event than the actual temporal order of these facts. The computer scientist should, therefore, pay particular emphasis to the format of the produced output of any explainable artificial intelligence system. The main message of the paper is that results from the social sciences must be regarded as a crucial foundation of any explainable artificial intelligence system.