58 Association of Research Libraries Research Library Issues 299 — 2019 the ability to collect data on most every aspect of a college student’s life, analysis of patterns can provide greater insight and predictions to assist students with their success. To be most effective, student data will need to be shared across university units in order to gain a greater understanding of the student’s performance and behaviors. Broadly sharing a student’s data across many members of the campus community can increase the risks of violating student privacy and regulatory protections such as FERPA and the Health Insurance Portability and Accountability Act of 1996 (HIPAA). Data security in these environments is also important and can impact how systems work with the data. The Library Integration in Institutional Learning Analytics project (LIILA) has documented their research and analysis of librarian involvement with learning analytics activities on campus.31 Librarians bring expertise in working with personally identifiable information, data privacy and security, informed consent, and access- controlled data storage. Leveraging this expertise can help universities adopt informed policies regarding the use of AI systems in making decisions that can have a significant impact on students. When black- box algorithms that are not transparent are using increasingly diverse data to make decisions and recommendations, having policies in place to enable more accountability will be important to explain how decisions are being made and confirm that bias is being held in check. Summary The computational ability to process massive amounts of data and detect patterns that can refine themselves over time enables a level of intelligence that humans cannot achieve due to cognitive limitations in processing truly large amounts of data and information. As Brundage and Bryson have observed, however, “Artificial intelligence is not necessarily similar or equivalent to human intelligence. In fact, because human intelligence keeps evolving (primarily culturally but even biologically) to meet the requirements of our animal lives and societies, it is unlikely that even if an AI was built to be exactly like human intelligence that it would stay that way for long.”32