Browsing by Author "Hoffström, Henriikka"

Goal of the study: Social support availability has been linked to less stress. Along with the human support there have been considerations of the possibility of artificial intelligence based conversational agents to offer support in stressful situations. According to prior research, conversational agents can influence people most effectively when it is known that they are controlled by human. It is not yet known how information about the source of social support affects stress related primary and secondary appraisals. The current study investigated how the information of the social support provider (support from a simulated research assistant/simulated conversational agent/no support at all) affects the primary and secondary appraisals related to challenging arithmetic task, whether the older groups’ reserve towards technology is reflected in the appraisals and how is the state of arousal and sense of control during the performance of a task, and how pleasant the task is perceived to be. Methods: There were 58 participants in the study (18-66 years, 70.7% women). Participants completed three test conditions, where they first chatted with a simulated research assistant or conversational agent, and in the control situation with no one. Questions about primary and secondary appraisals were answered in the chat. After that, they moved on to perform the arithmetic task chosen as a stress stimulus. Results: Source of social support predicted both primary and secondary appraisals. Primary appraisals were smaller and secondary appraisals were larger when support was received from a simulated research assistant compared to the situation where support was received from a simulated conversational agent. Age group and experimental condition did not have an interaction effect regarding appraisals, although secondary appraisals were greater in younger group in each experimental condition. Participants felt less pleasure in the test situation after the conversation with research assistant compared to the situation with the conversational agent. On the other hand, the state of arousal and control did not differ in these situations. Conclusions: Based on the results, it can be stated that the information of support from a human is associated to lower stress in a potentially stressful situation, while the support from a conversational agent is not seen as a resource. The most important thing in terms of social influence can be considered the information that you are talking with a human and not with an artificial intelligence. Older people's reservations about technology did not emerge in this study.