Artificial Intelligence (AI) agents that learn from humans are a known branch of AI research as it can achieve human-like decision making in safety-critical domains like autonomous driving. It is a plausible requirement that the behaviour of these agents is interpretable, so the user can understand the reasons that drive the decisions of AI systems. Global interpretations can help users understand what drives an AI model. This position paper explores the current state of interpretability in imitation learning. Algorithms for imitation learning belong to one of the two categories: behaviour cloning or inverse reinforcement learning. We determined the differences between the two categories with regard to the interpretation of the algorithms. Our analysis showed that the reward function of inverse reinforcement learning is suitable for global interpretations of the model. Global interpretations are less frequently used for behaviour cloning as they have no reward function.