The effect of variation in repetitive utterances on users by a chat-oriented communication robot

Purpose Robots with certain communication abilities through AI have been shown to improve cognitive functions in elderly individuals living alone [Tanaka et al., 2012]. However, for the effective deployment of communication robots in practical settings, it is crucial for the elderly to use them continuously. Repetitive utterances by communication robots have been suggested as a strategy to encourage users to continue engaging in the dialogue [Yang & Kikuchi, 2021]. Based on the above, this study focuses on the repetitive utterances by communication robots and examines the effect of variation in repetitive utterances on users’ engagement in the dialogue. Method In this study, based on the analysis of chat-oriented dialogue corpus between humans, the variation of repetitive utterances was expressed by the interjections or backchannels that co-occur with the repeated words. Subsequently, based on the OpenAI GPT-4 model, we constructed a chat-oriented communication robot that automatically generates repetitive utterances. We conducted a robot dialogue experiment (Figure 1) with 41 university (graduate) students, dividing them into two groups: one with no variation in repetitive utterances (No Variation Group) and one with sufficiently high variation in repetitive utterances (Variation Group). Results and Discussion Without considering the characteristics of the subjects, no statistically significant difference was observed in the ratings of the ‘No Variation Group’ and the ‘Variation Group’ regarding users’ engagement. However, when dividing the subjects based on their “negative attitude toward emotions in interaction with robots”(NARS-S3) [Nomura, Kanda & Suzuki, 2006], a significantly higher evaluation of dialogue continuity was observed in the ‘Variation Group’ among those with high NARS-S3 scores (n = 21, p = 0.044, r = 0.440). In the future, we plan to verify the effectiveness of repetitive utterances by communication robots targeting elderly people with dementia.