Explainability for AI and Law in the Wild

Explainability has recently come to the fore of concerns about deployment of AI in real world domains. The recent advances seen in data-driven techniques have brought AI research into mainstream applications, yet specific domains have different user needs and acceptance criteria for the deployment of various AI techniques. A key concern for AI and Law applications is the ability of predictive tools to provide a suitable explanation for the conclusions they draw, as is expected from humans who currently execute tasks involving legal judgment. This talk will provide a survey of some of the key issues relevant to the topic of explainability in AI and law research today, and then an overview will be presented of a body of work produced recently that is aimed at providing explainable decision support for legal case-based reasoning. A particular focus will be placed on the deployment of these methods on real world cases to demonstrate the accuracy of their conclusions and their explanation features, both of which are needed to ensure that trust is engendered in the AI-based tools that are being built to provide assistance to legal professionals.