Jönköping University

EXPLAINABLE ARTIFICIAL INTELLIGENCE (XAI) Artificial Intelligence is increasingly playing an integral role in our daily activities. Moreover, AI based solutions are used more and more in areas such as criminal justice, healthcare, and education, and therefore, their impact is high. The dominant role played by AI models in these domains has led to a growing concern regarding potential bias, and a demand for model transparency and interpretability. Why did the system make this prediction? Do I trust it? What would happen if I change some parameter? As a consequence, AI researchers and practitioners have focused their attention on explainable AI to help them better trust and understand AI models. In this course, we present an introduction to Explainable Artificial Intelligence (XAI). We describe the challenges associated with the use of black-box models and how we can overcome such challenges using interpretable and explainable methods. Moreover, we study other aspects related to interacting with AI-based systems, for example, trust, acceptance, evaluation, and Fairness Accountability and Transparency-issues. The course is given in English and is targeted for working professionals in the industry. This course gives an introduction to Explainable AI (XAI), providing an overview of relevant concepts such as interpretability, transparency and black-box machine learning methods. The course provides an overview of state-of-the-art methods for generating explanations, and touches upon issues related to decision-support, human interaction with AI/intelligent systems and their evaluation. In summary, the Explainable AI course covers the following topics: Definitions and concepts such as black-box models, transparency, interpretable machine learning and explanations. Decision-making and decision support, Human-Computer Interaction (HCI) and AI. Explainable AI. Methods for Explainable AI. Applications and examples. Trust and acceptance. Ethical, legal and social issues of explainable AI. Evaluation methods and metrics. After a successful course, you will be able to: Show familiarity with concepts within Explainable AI and interpretable machine learning. Demonstrate comprehension of current techniques for generating explanations from black-box machine learning methods. Demonstrate comprehension of current ethical, social and legal challenges related to Explainable AI. Demonstrate the ability to select and assess Explainable AI methods. Demonstrate the ability to review, present and critically assess state-of-the-art papers in relevant areas within Explainable AI.