Chatbots can be useful in many different contexts including research, teaching and learning, pastoral care and student support, professional services and operations, or libraries.
By reading this guide you will be able to:
- Identify potential use cases for chatbots
- Begin to have meaningful conversations about whether to adopt chatbots
There are two distinct ways to interpret the meaning behind user utterances (a written or spoken input), which results in two different categories of chatbots.
In one set of cases, a user will want to gain information from a chatbot, for example, by asking it questions.
To meet users’ needs, the chatbot will need to interpret the meaning of the question being asked so that it can select what information to give in response.
This is the underlying logic of how many domestic digital assistants work.
Imagine a user who wants to know how much time they need to hard-boil an egg. They may try to get this information by asking their digital assistant an appropriately phrased question:
"How long do hard-boiled eggs take?”
Aided by AI, the digital assistant’s role is to establish what information the user wants, and then to return the relevant information.
Here, the digital assistant takes the user’s utterance as an input and provides a response as an output.
In other cases, chatbots can be used to complete tasks that involve seeking information from a user.
In these cases, it tends to be the chatbots asking the questions, rather than the users.
Consider, this time, a domestic digital assistant, the functionalities of which include ordering shopping. In order to make appropriate purchases the digital assistant may need input from the user, which it could get by asking questions like:
“What meals are you planning to eat this week?”
“Have you used up all the onions you had ordered last week?”
By processing the user’s responses, the digital assistant should be able to establish which purchases to make.
There are other ways to categorise chatbots – for example, some chatbots are integrated with sources of data so that they can personalise responses, whereas others are not.
This categorisation is independent of whether a chatbot is information-based or task oriented. With a personalised and information-based chatbot, for example, a user might ask a question like:
“What lesson do I have next?”
and receive an answer in the form of:
“Your next lesson is Chemistry. It starts at 11am and is in room SC 12.”
To give this answer, the chatbot will have known:
- The identity of the student asking the question (the chatbot could have been accessed through a single sign on)
- The students’ timetable (the chatbot could have access to the information management system)
- The time of day (so that it can identify the next lesson)
With a similar information-based chatbot that is not personalised, the chatbot might give the response to the same question as:
“You can find out what your next lesson is by accessing your timetable, which is available here (link). Note, you will need to sign in to access your timetable.”
With this example, the chatbot does not need to know the identity of the student asking the question, nor does it need direct access to an information management system.
Teaching and Learning
For a maths course, there are numerous formulae that students need to use, however, they do not have to memorise them.
Rather than requiring students to read a lengthy formula sheet, the maths department develops a chatbot that allows students to ask, for instance:
“What is the formula for adding variances?”
and get the relevant formula in response.
The finance department has found that they continually receive similar queries about the expenses system.
They identify the most commonly asked questions, and build a chatbot that can respond to these questions - which, they hope will free up time for the team.
The pastoral team find that at the beginning of the academic year, students often come to them asking the same kinds of questions. This leads to the team having less time to support student wellbeing.
They deploy a chatbot that can respond to students’ frequently asked questions, and this frees up the team’s time to focus on more pressing pastoral issues.
Teaching and Learning
A particular multi-disciplinary course appeals to students from a range of academic backgrounds.
In order to find out more about the experiences, strengths and areas of development for learners, the course lead develops a chatbot that can source relevant information from students by asking questions and interpreting student answers.
The human resources department wants to make sure that employees understand the benefits and rewards they are entitled to, which vary depending on an employee’s circumstances.
The chatbot they decided to deploy can, for instance, ask questions of users to find out if they are entitled to free eye tests.
The careers and development department has judged that students would benefit from independent practice in between mock interview sessions so they can build up their skills in this area.
They deploy a chatbot that asks students interview questions and adapts follow-up questions based on their answers.
Ada, Bolton College
Bolton College uses Ada to answer both staff and students’ questions about college life. The chatbot is integrated with other platforms used by Bolton College, such as its information management system, which means that the chatbot can provide answers that are specific to any given user. A student, for instance, can get information about their timetable. Meanwhile, a teacher can get information about attendance figures in one of their classes.
Jisc’s chatbot pilot
Inspired by Bolton College’s successes (see above), Jisc’s national centre for AI launched a pilot, in which four UK colleges are using a chatbot. The chatbot that is being piloted was developed in-house by Jisc, taking inspiration from Ada’s design.
The four pilot colleges have used the chatbot in an equivalent way to Bolton. Of note, one of the colleges focused on using their chatbot to respond to queries about student funding, in order to lessen the demands on this team without compromising the service provided to students.
Taylor, The Open University
Designed to enable a dialogue with students who want to disclose a disability, Taylor serves as an alternative to the process of filling in forms, whilst also giving students the opportunity to ask questions about what studying at The Open University entails.
Beacon, Staffordshire University
Beacon serves a variety of purposes. As a digital coach, it helps students to structure their independent study time, whilst also motivating them to achieve key milestones. As a personal assistant, Beacon can order new ID cards for students or organise their council tax submissions. Meanwhile, through the use of natural language processing, Beacon can also recognise and respond to student questions on topics such as mental health and safeguarding.
- Guide: Preparing for chatbots by engaging key internal stakeholders
- Guide: Developing high-quality question-and-answer sets for chatbots
- Jisc’s blog on how digital assistants are promoting enhanced accessibility looks at how the Open University are using chatbots to support accessibility.
- Artificial intelligence (AI) in tertiary education, the national centre for AI’s 2022 report, includes information on chatbot functionality and use cases, and gives examples of chatbot use in education.