Evaluating chatbots’ adherence to the integrated model of advice giving: presence and sequential structuring of supportive acts in human-AI interactions

Incorporating the Integrated Model of Advice Giving (Feng [2009]. “Testing an Integrated Model of Advice Giving in Supportive Interactions.” Human Communication Research 35 (1): 115–129. ), this study investigates how the presence and sequence of supportive acts engaged by chatbots when offering advice affect user’s responses in terms of intentions to implement the advice and perceived facilitation of coping. In a between-subjects online experiment, participants (N = 1069) read simulated human-chatbot conversations that varied in the presence and sequence of supportive acts in an advice-giving context. Results show that problem inquiry and analysis significantly improved user’s responses, while emotional support only enhanced perceived facilitation of coping. The sequence in which advice was offered (e.g. before or after provision of emotional support) did not affect the user’s responses. Both perceived affective and instrumental communication skills of chatbots as well as valence of expectancy violation significantly mediated the effects of presence of emotional support on user’s responses, whereas problem inquiry and analysis influenced user responses only through perceived instrumental skills of chatbots.