The AA carries out a comprehensive Member survey programme using conventional sampling, questionnaires and analysis techniques. This, however, requires AA staff to identify topics of importance (usually derived from Government initiatives) as the subject of the survey questions. Beca presented an opportunity to work with the IBM Watson artificial intelligence (AI) platform to develop a chatbot survey tool which would allow Members to raise their own topics of concern.
Stage One of this project examined the use of artificial intelligence as a means to fill a gap in knowledge around AA Members’ views.
The AI tool used in this research was initially developed by Beca in early 2019 to enable a new and innovative method for engaging with people on transport issues. A key feature of the tool is the ability to use AI to process conversation transcripts and identify key issues.
For this research project, the tool was branded “Indicate” and given a “personality” and “tone” appropriate for conversations with AA Members. The tool was programmed with the purpose of finding out what AA Members think about transport issues.
The first stage of this project involved many trials, which began with very small groups of AA staff, who gave feedback on how user-friendly the tool was. Beca responded to this feedback by making refinements and the trial groups grew larger during 2020.
Once confidence had grown in the "Indicate" tool, an invitation was sent to AA Members as part of an email newsletter to experiment with it.
This larger trial showed that "Indicate" could gather opinions from AA Members who had a conversation with it, and this information could deepen our understanding of attitudes to, and experiences with, transport.
Why is the AA interested?
The AA is aware of potential limitations of questionnaire survey techniques. AI presents an opportunity to capture people's views differently, which could reveal information not previously conceived by the transport sector.
Stage One - Results
- Matthew Ensor - Beca Ltd
- Simon Douglas - AA Research Foundation
- Peter King - AA Research Foundation
Page last updated: May 2022