The advancement of artificial intelligence and its diverse applications have attracted great interests from several research scholars examining communication between humans and chatbots. Many have explained the motivations of chatbot use, however, less emphasis has been given to myriad customer experiential needs for the interaction with this AI-powered technology. Therefore, in the light of U&G theory, this thesis presents a conceptual framework for customer experience as the needs for chatbot use. SOR model was also adopted to establish the links among customer experience as organisms (utilitarian – perceived informativeness and credibility; hedonic – perceived transparency, enjoyment and engagement; anthropomorphism – mindful and mindless anthropomorphism; and social presence), their primary stimuli (functionality, communication style similarity and aesthetics) and responses (customer satisfaction and reuse intention). 417 convenient samples were collected from the UK where chatbots are widely used. Results from SEM revealed that perceived credibility, informativeness, enjoyment, functionality and communication style similarity hold the keys for customer satisfaction, while effects of anthropomorphism and social presence are not significant in this research setting. This study has enriched U&G theory by addressing customer experience as ones’ motivations to use chatbots. Also, it has presented a fresh understanding of customer experience in chatbot context by considering utilitarian, hedonic, anthropomorphism and social presence as dimensions of customer experience rather than merely measuring how customers think, feel, sense, act and relate. In addition, practical implications and direction for future research are discussed.

Emotion AI är ett forskningsområde som syftar till att teknik ska kunna känna igen och reagera på mänskliga känslotillstånd, vilket kan implementeras i form av chattbotar. Chattbotens roll är idag framträdande inom kundtjänstområdet och används för att stärka kundservice. Tidigare forskning visar att det råder brist av kvalitativa studier om hur chattbotar responderar på och hanterar känslor som uppstår när människor interagerar med chattbotar i vardagen. Syftet med studien är därav att skapa förståelse för hur känslor hanteras av chattbotar och vilka upplevelser som genereras vid kundtjänstinteraktion mellan människa och chattbot.  Studien antog en kvalitativ ansats och inleddes genom en förstudie som identifierade fyra chattbotar, vilka användes vidare till intervjustudien. Datainsamlingen har genomförts med hjälp av semistrukturerade intervjuer och kompletterades med observation. Analysen identifierade fyra typer av upplevelser av chattbotarnas förmåga att respondera på känslointeraktion: bekräftelse, missförstånd, problemlösning och ideal chattbot. Studien diskuterar hur chattbotar påverkar människans känsloupplevelse och hur chattbotar bör designas i kundtjänstinteraktion. Slutsatsen är att det inte finns behov att utveckla chattbotar med väldigt komplex känsloförmåga och känslomässig förståelse har visats vara sekundärt i kundtjänstinteraktion. Dock bör chattbotar designas med en enklare form av känslomässig förmåga eftersom det kan skapa mervärde genom att kunderna känner sig förstådda. / Emotion AI is a research area aimed at technology recognizing and reacting to human emotions, which can be implemented in the form of chatbots. The chatbot's role is today prominent in the field of customer service and is used to strengthen customer service. Previous research shows that there is a lack of qualitative studies on how chatbots respond to and manage emotions that arise when people interact with chatbots in their everyday life. The purpose of the study is therefore to create an understanding of how emotions are handled by chatbots and what experiences are generated during customer service interaction between humans and chatbots.  The study adopted a qualitative approach and was initiated through a feasibility study that identified four chatbots, which were used for the interview study. The data collection was carried out using semi-structured interviews and supplemented by observation. The analysis identified four types of experiences of chatbots' ability to respond to emotional interaction: affirmation, misunderstanding, problem-solving, and ideal chatbot. The study discusses how chatbots affect the human emotional experience and how chatbots should be designed for customer service interaction. The bottom line is that there is no need to develop chatbots with very complex emotional abilities and emotional understanding has been shown to be secondary in the customer service interaction. However, chatbots should be designed with a simpler form of emotional ability as it can create added value in the form of making customers feel understood.