The aviation industry operates within a highly critical and regulated context, where safety and reliability are paramount. As Natural Language Processing (NLP) systems become increasingly integrated into such domains, ensuring their trustworthiness and transparency is essential. This paper addresses the importance of explainability (XAI) in critical sectors like aviation by studying NOTAMs (Notice to Airmen), a core component of aviation communication. We provide a comprehensive overview of XAI methods applied to NLP classification task, proposing a categorization framework tailored to practical needs in critical applications. We also propose a new method to create aggregated explanations from local attributions. Using real-world examples, we demonstrate how XAI can uncover biases in models and datasets, leading to actionable insights for improving both. This work highlights the role of XAI in building safer and more robust NLP systems for critical sectors and also shows that academic efforts must be pursued to achieve trust in models and XAI itself.

When automating plan generation for a real-world sequential decision problem, the goal is often not to replace the human planner, but to facilitate an iterative reasoning and elicitation process, where the human's role is to guide the AI planner according to their preferences and expertise. In this context, explanations that respond to users' questions are crucial to improve their understanding of potential solutions and increase their trust in the system. To enable natural interaction with such a system, we present a multi-agent Large Language Model (LLM) architecture that is agnostic to the explanation framework and enables user- and context-dependent interactive explanations. We also describe an instantiation of this framework for goal-conflict explanations, which we use to conduct a user study comparing the LLM-powered interaction with a baseline template-based explanation interface.

ConSim is a metric for concept-based explanations based on simulatability and user-llms. It shows consistent methods ranking across datasets, models, and user-llms. Furthermore, it correlates with faithfulness and complexity.