38 Association of Research Libraries Research Library Issues 299 — 2019 It is concerning that these innovations are occurring outside the field of academic librarianship and with little or no involvement of library expertise. If libraries are to shape AI development and embed values such as explainability in these tools and services, it is essential that the challenges voiced by Lankes, Bourg, and Coleman be acknowledged, accepted, and acted upon. In addition to the focus on innovation in tools and services, academic libraries can further XAI through such avenues as public policy and algorithmic literacy. Public Policy A key XAI strategy is to use authorizations, such as legislation, regulation, and audit, as governance methods to support, or even require, explainability. Despite widespread concerns about algorithmic decision-making with respect to bias, discrimination, and unfairness, this is an area that is largely unregulated in Canada and the United States. The AI public policy landscape is nascent. Some have argued for a “regulatory lag” to allow more clarity on how AI will evolve, while others more cynically dismiss all regulations as solving “yesterday’s challenges” and impeding innovation in a globally competitive “AI race.” While premature and reactive regulation is undesirable, neither is an environment where abuses, harms, and predatory practices are allowed to exist. Research libraries, through organizations such as the Association of Research Libraries and the Canadian Association of Research Libraries, have a strong interest in influencing public policy and have achieved substantial successes in this area, even if only in raising public awareness. While it is argued that blanket AI regulation will be less effective than application-specific regulation (for example, If libraries are to shape AI development and embed values such as explainability in these tools and services, it is essential that the challenges voiced by Lankes, Bourg, and Coleman be acknowledged, accepted, and acted upon.