Doris’ research has recently looked at how the prompt:response turns of a commercial chatbot can be improved to make the bot easier to interact with and customers more likely to complete a sales interaction to completion.
This research was born out of a consultancy project with Account Management Limited Online, a commercial chatbot development company in Surrey Business Park. Company directors initially contacted the university because feedback from users indicated that customers though the bot was ponderous, annoying, underwhelming or even simply boring.
Doris and her co-author Jenny Lynden, a psychologist, drew on theories and concepts derived from linguistics (e.g. politeness, repair, adjacency pair) to investigate, using micro-analysis of language, how the language used for the prompts and responses supported affiliation between the user and the bot and, ultimately, trust. The analysis led to AMO reprogramming the prompt:response turns of their bot which has now been deployed to a number of their clients.
Doris’ research, which uses a language-led approach can be applied to a wide range of situations in which humans interact with technology, for example apps which support individuals’ health and wellbeing. Doris’ argues that such technologies need to be carefully designed to create trust, and to speak to different audiences whose requirements may not be all the same.
By Dr Doris Dippold, School of Literature and Languages.
