Lately, I’ve been encountering more and more books about AI and its impact in the public and private sectors. Works such as The Big Nine: How the Tech Titans and Their Thinking Machines Could Warp Humanity by Amy Webb, How to Speak Machine: Computational Thinking for the Rest of Us by John Maeda, and The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power by Shoshana Zuboff provide deep analyses of AI technologies in our community and the opportunities and challenges that AI presents to our world.
In addition, my email inbox has been filled with updates and reports coming from various library organizations. This year, the Association of Research Libraries (ARL) published a report on the ethics of AI and implications for research libraries (publications.arl.org/rli299). IFLA released a statement concerning the ethical developments and use of AI based on human rights. Anyone can view and comment on this document (“The Human and the Algorithm: Response to Council of Europe’s Draft Recommendation on Human Right Impacts of Algorithmic Systems”; www.ifla.org/node/92559?fbclid=IwAR0LL0YD8sJHR7tH7r6V6bMWMod8yMKzM4Dy1LW0HE_VpCgr7g1ftJlN4e0).
From ALA, Immediate Past President Loida Garcia-Febo shared several interesting examples of how libraries are integrating AI into their workflows, from discovering to accessing information, but encourages more conversations on AI (“How Libraries Are Starting to Apply Artificial Intelligence in Their Work”; americanlibrariesmagazine.org/2019/03/01/exploring-ai). More reports on this topic are expected to be released within the next year.
Go to any library/information science-related conference today, and you’ll find one, two, or multiple sessions on AI and its impact on services, resources, and the ecology of information. It is not surprising that while AI is very present in the literature and in our learning spaces, many folks may still be left wondering how AI will impact the profession. As noted in previous reports, libraries have started integrating AI into their services. For example, the University of Rhode Island (URI) Libraries has the Artificial Intelligence Lab to support multidisciplinary research and collaboration, as described by URI’s associate professor Bohyun Kim in the January 2019 Library Technology Reports (journals.ala.org/index.php/ltr/article/view/6910). [Bohyun Kim is a columnist for Online Searcher. —Ed.]
In August 2019, I attended the IFLA WLIC (World Library and Information Conference) held in Athens, and in one jam-packed session on AI and data mining, speakers posed how AI has worked well with data mining while also cautioning the concerns of bias, ethics, and violations of user privacy (“Yewno: Transforming Data into Information, Transforming Information into Knowledge”; library.ifla.org/view/conferences/2019/2019-08-26/1029.html). Philip Schreur from Stanford University argued that AI can be effectively used to suggest subject terminology in the cataloging of electronic theses and dissertations (ETDs). And with advances in facial recognition technology (assuming issues of privacy can be addressed), perhaps search interfaces can automatically ad just their behaviors based on individual patrons.
Major concerns include data privacy, biases, and the lack of regulation and accountability under such algorithmic systems. In September 2019, at the Association of Library and Information Sciences Education (ALISE) conference in Knoxville, Tenn., a session focused on data ethics and policy and how the LIS curriculum should incorporate data ethics into classes to prepare students for data-centric careers and to develop an understanding on ethics and policy issues of data collection practices by AI technologies on communities, particularly vulnerable ones. Another relevant session discussed how some LIS graduate programs are undergoing organizational and strategic changes and are forming new schools by partnering with computer science and informatics departments. This type of academic structure can center LIS education with disciplines that focus on AI, thus potentially enabling LIS students to gain more fluency in AI systems.
One big question that seems to rise from such conversations in our profession is whether AI will be replacing jobs, specifically, the roles of library/information workers. Rather, the question should be, “ How will AI be enhancing and evolving our current job roles?”
This inevitable reality requires us to recalibrate and revisit our own professional practices and tasks. What and how can we develop our work alongside AI as a partner? AI can be programmed to do repetitive work at an accelerated rate or to conduct deep and machine learning (ML) through sifting data. With the arrival of 5G (the fifth generation of mobile phone communication standards), machine-to-machine interactions will swiftly bring new types of analyses and interactions that have not been seen or performed before. ML has been programmed to conduct Big Data collection at any convenience.
As library/information sciences professionals, we need to deeply reflect on the impact of these types of activities that have permeated from the private sector and will bring on all sorts of changes and challenges. As a matter of fact, we need to be focused on including such services, but also need to craft guidelines and policies for regulation, respect, and adherence to data privacy concerns, particularly from our users, to minimize the biases generated by AI technologies.
Obviously, these are basic necessities and for the long term. Collectively, we need to strategize our workflows that will be impacted by AI. One area that certainly will be affected by AI is technical services. For instance, the role of metadata will only increase in importance, but how it is created and maintained will continue to evolve. Currently, there are human-centered AI projects in which AI is an active partner with catalogers to create new ways of approaching information access that would have been impossible by AI or humans alone. By using well-known techniques, including named entity recognition on digital text, entities such as people, places, or events can be automatically extracted and identifiers assigned via Wikidata, providing deep access to large bodies of resources too expensive to catalog by humans alone.
Partnership is key. We need to continue conversing, identifying, and minimizing AI’s biases and flaws while recognizing AI as a partner in fostering discoveries and access to information. AI is here to stay.