Beyond Turn-taking: Introducing Text-based Overlap into Human-LLM Interactions

TL;DR

This research proposes a novel approach that incorporates overlapping messages, mirroring natural human conversations, and provides recommendations for implementing overlap-capable AI interactions to enhance the fluidity and engagement of text-based conversations.

Abstract

Traditional text-based human-AI interactions often adhere to a strict turn-taking approach. In this research, we propose a novel approach that incorporates overlapping messages, mirroring natural human conversations. Through a formative study, we observed that even in text-based contexts, users instinctively engage in overlapping behaviors like "A: Today I went to-" "B: yeah." To capitalize on these insights, we developed OverlapBot, a prototype chatbot where both AI and users can initiate overlapping. Our user study revealed that OverlapBot was perceived as more communicative and immersive than traditional turn-taking chatbot, fostering faster and more natural interactions. Our findings contribute to the understanding of design space for overlapping interactions. We also provide recommendations for implementing overlap-capable AI interactions to enhance the fluidity and engagement of text-based conversations.

Figures (8)

• Figure 1: Current turn-taking chat with an LLM. Motivated by the absence of overlap points, we identified a new opportunity to design overlap-capable text-based interactions.
• Figure 2: Visually changed UI from typing status to sent status.
• Figure 3: Examples of the types of overlap by OverlapBot. While the user is typing, OverlapBot can provide listener cues indicating attention (Backchanneling) or generate a response before the user finishes typing (Preemptive Answering).
• Figure 4: The chatbot regenerates an answer based on the interruption, presented in a timely order from (a) to (c): Before the interruption (a), the chatbot is generating an answer about sentiment analysis. The user interrupts by asking about machine translation (b). After the interruption (c), the chatbot generates a new answer about machine translation.
• Figure 5: First new human-LLM interaction: The user checks OverlapBot's active listening through its preemptive answering.