“From simply one search”: Exploring the effects of task complexity on the use of GenAI in search

Generative AI (GenAI) is increasingly shifting the landscape of academia. With improvements in GenAI models and wider availability of GenAI tools, such as ChatGPT, the needs and behaviors of users are changing. For the field of Information Retrieval (IR), these changes require researchers to investigate what this might mean for user-centric designs and support tools, particularly for the support of complex and exploratory tasks. This thesis explores how task complexity may influence participants’ perceptions of GenAI in search. Through a within-subjects study conducted with 16 participants, all undergraduate students, findings show that there were no significant differences of perceptions between tasks of different levels of complexity. Participants indicated that they were familiar with using GenAI tools and reported enjoying the ease using GenAI afforded them. Despite finding no significant differences from the manipulation of task complexity, there was a significant difference in participants’ perception of post-session success based on the order of completion for tasks. In this study, additional data was collected from think-aloud procedures and interaction behaviors, which will be analyzed in future work.