When is my next assignment due? How much leave do I have left this year? When is the next bus into town?
One thing these questions have in common is they can all be answered relatively easily by a chatbot. In some institutions, they already are.
These examples come from a project my colleague Paul Hopkins and I have been conducting, co-sponsored by Jisc and UCISA, to survey the opportunities that chatbots offer and what further and higher education institutions can do – and are doing – to exploit these.
Our chatbots and digital assistants report outlines maturity models for chatbots, the current state-of-the-art in respect of some of the major vendors’ technologies and gives some tips and pointers for getting started in this field.
Chatbots have some real positives:
- They are scalable when a cloud service is used (which is the most common approach) so can handle seasonal and unexpected peaks
- They are consistent
- They are available round the clock
- Their impersonality can be an advantage – a chatbot will not judge anyone if they are struggling
If they are integrated well into other systems and have access to a good, well-ordered source of data, they can offer a personalised and context-sensitive service, for example, “Your assignment is due in next Wednesday, but remember you also have deadlines on Thursday and Friday.
Of course, there are aspects to beware of. Some of the key lessons we’ve learnt so far include that, when it comes to developing a full-scale service that can operate effectively with a wide audience, careful preparation and design is needed.
It’s also important to have a well-ordered knowledge base – the corpus of knowledge that underpins the chatbot. This might require some preparatory work for some institutions. The role of service owners, outside of IT, is also crucial, and getting them involved in assessing and developing chatbots is a good step to take early on.
The project continues into 2019 and I’d be very keen to hear from institutions that are working in this field or considering doing so.
Direct to Complete Blog Post
Direct to Full Text Report (20 pages; PDF)