Explainable Artificial Intelligence (XAI): Current Approaches and Paths to the Future

There is a resurgence of interest in artificial intelligence (AI) applications in biomedical domains. There is a concomitant interest in how such applications reach a conclusion, such as a prediction or classification. Given that AI systems in biomedicine can affect a user’s decision about providing patient care or choosing a particular algorithm for mining data, it is critically important for informaticians and computer scientists to create explainable AI systems to address this. This panel will review the history of explainability in AI, and introduce four areas in which AI is developed, used, and evaluated.