The dominant patterns of librarianship were shaped decades ago, before the widespread adoption of the great two innovations of the 21st century: the Internet and Artificial Intelligence (AI). The old patterns were all based on two (now dated) ideas: local holdings and third-party indexing. Librarians would collect as much content as possible and would rely upon third parties to provide indexing for their journal and multi-authored works. That work was then stored on a shelf somewhere until a library patron stumbled upon a reference to a chapter or article of interest through a scholarly citation, a printed index or a pay-walled digital index. Once the patron identified a chapter or article that interested them, the patron would go to the library to retrieve the content, or if the content was not in the library's local holdings, the patron would submit an InterLibrary loan (ILL) request. These processes were slow, inefficient, expensive and resource-intensive. At the Digital Theological Library, a library co-owned by over 80 institutions across the globe, we are trying to rethink every piece of these familiar processes. In a multi-year project, we and our third-party partners are using Artificial Intelligence to modernize the delightfully 20th-century approach outlined in the previous paragraph. Our project, which relies upon generous third-party funding, is proceeding in three stages:
- We are using AI to index our vast collection of journals and multi-authored books (the DTL owns over 1,000,000 print volumes);
- we plan to use AI to create abstracts of all of our already digitized content (currently nearly 100,000 individual volumes, and growing quickly);
- we hope to eventually create an ILL system which makes requesting chapters and articles both quick and easy; and
- depending upon how future legislation and/or legal decisions turn out, we hope to create a scholarly AI inquiry site for religious studies.
At present, we plan to make the metadata created in the first part of the project freely available to any interested party; we likewise plan to make the abstracts available via a free public website at no cost to users (thanks again to external funders). Since the first two parts of our four-part project are standard and long-established academic practices (indexing and abstracting), we foresee no copyright concerns. Although we would love to provide the chapters and articles to the public without costs, copyright concerns will require us to limit access to our content to libraries through InterLibrary Loan (which is, of course, another standard and well-established library practice). As further concessions to copyright concerns, we are concentrating our attention on content which cannot be licensed electronically and content which is out of commerce. The development of an AI-powered inquiry system is dependent upon future legal restrictions around training a Large Language Model with copyrighted material. In terms of sustainability, after digitization, we are able to store print volumes in extremely high-density conditions, averaging about 30 volumes per cubic foot. We preserve up to three print copies of each work, freeing small libraries of the need to preserve content which is low use. We believe that we are helping libraries engage in the one key practice which will reduce their carbon footprint more dramatically than any other single practice; we are helping libraries to reduce their physical footprint.