Development of an Explainability Scale to Evaluate Explainable Artificial Intelligence (XAI) Methods

Explainable Artificial Intelligence (XAI) is an area of research that develops methods and techniques to make the results of artificial intelligence understood by humans. In recent years, there has been an increased demand for XAI methods to be developed due to model architectures getting more complicated and government regulations requiring transparency in machine learning models. With this increased demand has come an increased need for instruments to evaluate XAI methods. However, there are few, if none, valid and reliable instruments that take into account human opinion and cover all aspects of explainability. Therefore, this study developed an objective, human-centred questionnaire to evaluate all types of XAI methods. This questionnaire consists of 15 items: 5 items asking about the user’s background information and 10 items evaluating the explainability of the XAI method which were based on the notions of explainability. An experiment was conducted (n = 38) which got participants to evaluate one of two XAI methods using the questionnaire. The results from this experiment were used for exploratory factor analysis which showed that the 10 items related to explainability constitute one factor (Cronbach’s α = 0.81). The results were also used to gather evidence of the questionnaire’s construct validity. It is concluded that this 15-item questionnaire has one factor, has acceptable validity and reliability, and can be used to evaluate and compare XAI methods