We live in a time of exponential growth and sharing of data. Younger generations seem more open to share their data behaviors with others, and they also expect customized services based on their needs and preferences. Some say that chatbots along with custom or smart portals can be tools for academic libraries to push, embed, integrate, and customize just-on-time scholarly resources to students. Question: What do you think is the most cost-effective and ethical use of AI-powered services in academic libraries?
Artificial intelligence, especially machine learning and algorithmic responses, do seem to promise interesting services in some library settings. In a purely commercial environment, these technologies make use of vast amounts of personal information gathered from online and in-person user behavior. Yet, despite such aggressive use of data, it does not seem that the AI-driven interfaces in social networks and consumer services are necessarily all that effective. No chatbot that I have ever interacted with has ever led me to a satisfactory solution of a problem. The state of the art in AI-driven advertising networks seems to be to recommend things already purchased. This skepticism informed by consumer services tempers my view of how well these technologies can be used in library services. Any artificial intelligence used the library ecosystem will need to have additional layers to handle user privacy and will especially need to be designed to eliminate the possibility of bias.
With those caveats, I do see some opportunities to use AI for academic library services. The commercial discovery services, for example, are increasingly using these technologies to help build their indexes and interfaces. The construction of discovery indexes, given their massive scale, has always depended on automated tools to ingest and normalize the ingestion of content resources. Additional layers of automation, including those based on machine learning, can enhance metadata to improve the quality of search results. Increasingly discovery interfaces can rely on clues regarding the search context to guide the researcher toward more relevant results. The April 2017 issue of Smart Libraries Newsletter featured Yewno, a discovery service based on machine learning, concept extraction, and other elements of artificial intelligence.
I also anticipate the enhancement of library chat services with capabilities to provide some initial guidance for researchers. This level of automation has been around for a long time to help guide users toward specialized databases or to make contact with the appropriate subject specialist or librarian.
AI-based services can operate at a scale not possible through human operations. Commercial organizations implement these technologies to reduce or eliminate the costs of human expertise in their customer service operations. When implementing these technologies, it seems important to meet user expectations in solving routine questions, but to also leverage the human expertise that is the hallmark of library services.