Chatbots in Libraries

Chatbots automated conversation systems – have several advantages. They are available 24/7 and are consistent and patient when answering queries. They remove the requirement to navigate many different sources of information. In some cases, they are less intimidating for users and fit trends of communication involving smartphones or tablets 

Over the last few months, I have been looking into understanding the role of chatbots in the library. I’ve had conversations with a number of librarians (mainly in HE) and students to understand their views on how students could be supported by a chatbot service when using the library. In this blogpost I’m summarising what I have learned so far from these conversations.  

Chatbots are not new to HE or FE and the progress of chatbot technology made it possible to use them more widely in an education context. There are various tools, frameworks and development platforms to easily build them and there are few places in the university and college ecosystem where chatbots can’t be deployed 

Chatbots could potentially have many applications across the institution, from answering questions during student recruitment and enrolment to services across campus such as libraries or student support. This range is demonstrated by some of the chatbot services already rolled out by universities and colleges. In 2018, Jisc published the report ‘chatbots and digital assistants’ getting started in HE and FE which outlines maturity models for chatbots, describes popular vendor platforms, gives some tips and pointers for getting started in this field and gives a list of example developments in the HE and FE sector which have certainly grown since the report was published.

Some chatbots in libraries have existed for more than 10 years. Stella, for example supported users at Hamburg state and university library from 2004-2015. Some more recent examples include campuswide digital assistants which can also answer questions about the library (e.g. Beacon at Staffordshire University) or include library specific integrations (e.g. Deakin Genie at Deakin University). Some universities and colleges are starting to make plans for a library chatbot or are about to get started (e.g. Gower College, Swansea).

There are lots of library related questions that could be answered by a chatbot including questions where one quick answer is needed. In some cases, thousands of students have the same problem. For example, tool x doesn’t work with browser x; why can’t I access ebook x, what databases do you have on x etc. At peak times library staff answering questions at the other end of the chats systems which are often part of the library’s support services are acting like ‘chatbots. Also, many students work around the clock and often need answers to basic questions at any hour. Chatbots could also be of value to certain types of users such as distance learnersFor example, a distance learner I spoke to described how useful it would be to receive more immediate library support, as emailing the library or arranging a face-to-face meeting with a librarian takes a long time to resolve queries. 

Being able to get answers to simple questions such as when does the library open rather than having to search for this yourself seems to be reasonably useful. More mature chatbots driven by AI which can understand and answer a variety of questions could also help students with more complex tasks and answer questions in a context-based manner.

A scenario for example could be services to support students with searching for library resources which can be a complex and difficult task. Bizzy at the University of Oklahoma library, for example, is trained to help users to do a basic search for a book. It is changing previous processes of search and discovery as it can provide responses to questions asked using natural language rather than structured (author/title/subject) searches.

The librarians I spoke to emphasised that they would want to think carefully about the complexities and implications of integrating search functionality into a chatbot to ensure that it doesn’t interfere with the library’s role to support students to become digitally literate learners. In an early instance of Deakin Genie, (a chatbot at Deakin University in Australia) for example its interface attempted to simplify the search experience for students but it became clear quickly that in doing so it risked oversimplifying the academic search process. As a result of this Genie was amended to prevent this and to rather guide students on how to refine their search.  

It is also important that the service is implemented in a way so that it doesn’t affect library staffs’ connection with students. Face to face contact helps librarians to increase their understanding of the needs of students when using the library and contributes to a supportive environment.  

Another challenge for libraries is when they are a stakeholder as part of the development of a single campus wide chatbot or digital assistant – something our discussions suggested was a likely path for a number of universities and colleges. The library will need to decide how it would fit into the service – which might offer a central point of enquiry for students and which library systems it might be integrated with. Equally, if a library decides to develop its own chatbot it would want to avoid a situation where a student would need to use multiple chatbots to get answers to questions related to different types of services. This would result in a fragmented user experience.  

With current events driving a move to online delivery of courses, and the need to provide 24/7 support along with the intention to free up staff time from routine queries, chatbots seem to be a strong candidate to help address some of these challenges. At Jisc we are planning to keep an eye on developments in this space. During my research I didn’t come across any chatbots implemented by a UK university library though some institutional chatbots incorporated support for library services.
We’re keen to hear from any libraries implementing such a service, so please do get in touch.

See a list of resources and examples below.

Examples and resources related to libraries

CSGUK Conference 2019 in Cardiff
included discussion session on impact of future technologies on the library
(Also mentions the challenge of AI systems being able to handle Welsh).  

Library chatbot guide from San Jose State University in US 

Recording of OCLC webinar about ‘Deakin Genie’

-‘Bizzy’ at University of Oklahoma

FE example
Gower College Swansea chatbot project
This project aims to support vocational, distance and work-based learners in their chosen course of study via a library / research specific chat bot.  Gower College, Swansea aims to create a chatbot that is specifically designed to support students with their research and study skills needs.  The overall aim seeks to address the issue of parity of service for our students who cannot physically come into college and students who need assistance outside business hours.  The chatbot will address research, study and skills needs and be available across all devices and platforms. This project will reach 9,300 vocational learners across Swansea and the UK who are based within Gower College Swansea and the potential to reach vocational learners across Wales and the UK. 

More general resources and example

Jisc report (2018) Chatbots and digital assistants in HE and FE, how to get started

Southbank University chatbot 

-Article about ‘Becky’ (Leeds Beckett University) 

-‘Beacon’ at Staffordshire University 

FE examples  

-Aftab Hussain’s blog which includes information about Bolton college’s chatbot ‘Ada’: 

The Welsh Government provided funding under the banner of Digital 2030 to  

-Cardiff and Vale College for a Personalised Learning Bot for Maths and English GCSE
A personalised learning bot that will provide learners with specific content in their areas for development and
-a chatbot at Bridgend College

Print Friendly, PDF & Email

Leave a Reply

Your email address will not be published. Required fields are marked *