The Effect of Text Movement on Eye Movements in Generative AI Chatbots

Abstract

Background Text generating artificial intelligence(AI) chatbots are gaining attention as a new information exploration service. The way participants receive information has expanded through interaction and communication with these chatbots. This study aims to investigate the influence of the nonverbal element of text movement on informative conversations between text generating AI chatbots and conversation participants through eye-tracking. Methods The key attributes were derived through a theoretical review of movement, and these were applied to the conversational interface of the text-generating AI chatbot, classifying them into four types. Eye movement data was collected through eye-tracking while 36 experimental participants read the conversational text generated by the chatbot. In the analysis, differences were identified through repeated measures analysis of variance (ANOVA), Bonferroni post-hoc analysis, and independent samples t-test. Results The direction and focus, the elements of text movement occurring from the bottom, were relatively high in terms first fixation duration, number of whole fixations, and saccade amplitude. This appeared in consecutive conversations with ChatGPT and Bing AI, affecting higher cognitive activities. Additionally, an increase in generating speed of conversational text influenced the differences between text movement types. Therefore, the movement elements must be combined on the top of the screen to reduce excessive cognitive activities. Conclusions In order to achieve effective visual communication with the use of text-generating AI chatbots, design should consider how the text movement impacts the user’s eye movement and cognition, which aims to enhance the comprehension of information while also limiting cognitive overload.