36 Association of Research Libraries Research Library Issues 299 — 2019 The efficacy, and likelihood, of legislation mandating explanatory AI is widely discussed among researchers. While US, and to a lesser extent Canadian, past practice signals a reluctance to legislate in these areas, the EU, France, and the United Kingdom are taking different and more proactive approaches as exemplified by the GDPR. As a result, in Canada and the US, the most common recommendation for AI oversight and authorization is the use of a regulatory agency. Such an agency would have legislated or delegated powers to investigate, certify, license, and arbitrate on matters relating to AI and algorithms, including their design, use, and effects. The breadth and depth of the responsibilities of these agencies varies by those promoting them and by the relevant jurisdiction. Specific suggestions for a public agency include a “neutral data arbiter” with investigative powers like the US Federal Trade Commission, a Food and Drug Administration “for algorithms,” a standing “Commission on Artificial Intelligence,” quasi- governmental agencies such as the Council of Europe, and a hybrid model combining certification and liability. There are few calls for an international regulatory agency despite the global reach for many, if not most, AI systems and services. XAI and Research Libraries Algorithmic decision-making is already pervasive in information tools and services acquired, provided, or developed by research libraries. Often the methods and processes of those tools and services are invisible or unacknowledged. If libraries are to trust the quality, value, and credibility of these innovations, it is important that they be explainable. David Lankes warns of a new digital divide with “a class of people who can use algorithms and a class used by algorithms,”19 and argues that “librarians need to become well versed in these technologies, and participate in their development, not simply dismiss them or hamper them. We must not only demonstrate flaws where they exist but be ready to offer up solutions. Solutions grounded in our values and in