At the Business & Entrepreneurial Economics (BEE) Conference in early June, David Dobrowsky, Head of the Bachelor’s program in Corporate Communication, presented the results of a study on the impact of emotionalized language on the perceived credibility of chatbots.
From June 1 to 4, 2023, the Business & Entrepreneurial Economics (BEE) Conference, organized by the University of Zagreb, took place for the 8th time. David Dobrowsky, Head of the Bachelor’s program in Corporate Communication, presented results from a study on the influence of the interaction style from chatbots on the affective perception of users. He had conducted the study with Senior Researcher Ilona Pezenka and Gerald Janous, Head of the Master’s program in Marketing & Sales Management, in cooperation with Haaga Helia University.
His talk dealt with the effect of emotionalized language on the perceived credibility of chatbots. He focused on two types of a retrieval-based chatbot. Here, the chatbot generates the content of its answers from a pool of predefined information:
- Task-oriented interaction style with factual and formal communication
- Conversational interaction style with a linguistically less formal communication, where the chatbot tries to establish a conversation by using emojis
In the empirical study, the probands interacted with each type of the chatbot. The results show that a conversational interaction style leads to an increased perceived credibility of the chatbot by the users.
David Dobrowsky also presented a research design that measures the impact of emotionalized language on the perceived credibility of generative chatbots. In contrast to retrieval-based chatbots, generative chatbots generate the content of the answers themselves, for example by using a language model (language model based chatbots). However, even with these chatbots, the information for the generated answers can be specified upfront.