Curiosity as Linguistic Intervention: Using LLM Tutoring Dialogues to Influence Exploratory Learning Behavior

The article introduces CURIOBOT, a framework leveraging Large Language Models (LLMs) to implement adaptive linguistic interventions based on Berlyne's collative variables to enhance exploratory learning behavior. Through 270 tutoring dialogues across various model families and domains, curiosity-oriented interventions resulted in up to 2.4 times more conversational turns, indicating a significant increase in learner engagement without altering tutor-side instructional quality. This research highlights the potential of LLM-mediated dialogue as a scalable tool for investigating the impact of language on exploratory cognition, which is crucial for developing more effective AI tutoring systems.