Conversational AI systems often generate responses containing unfamiliar or complex words, which can disrupt user understanding when users must ask follow-up questions for clarification. This work presents a conceptual interface design for inline lexical assistance in conversational AI interfaces, where users can directly select or hover over words within an AI-generated response to view their meanings and similar words without interrupting the conversation flow. The proposed approach focuses on user experience rather than model-level changes, enabling seamless comprehension support through contextual tooltips or pop-up elements embedded within the chat interface. This design aims to reduce cognitive load, minimize conversational disruption, and improve accessibility for users with diverse linguistic backgrounds. The paper discusses the motivation, interface workflow, and potential benefits of the approach, supported by illustrative interface mockups and a system-level flowchart. This work is intended as a preliminary design and concept exploration, serving as a foundation for future implementation and user evaluation studies.
Bharath Kusuma (Wed,) studied this question.