What Is Your Site's AI Chatbot for? Users Can't Tell

Why Read This from an HCI Perspective

This article clearly demonstrates from an HCI perspective why AI chatbots on websites are not being used as much as expected, and under what conditions they actually provide value. It empirically addresses issues such as discoverability, information scent, cognitive load, and redundancy with existing search and filtering, making it highly relevant for both UX practitioners and researchers. It reaffirms the importance of defining user problems first before attaching AI functionalities.

CIT's Commentary

From a CIT perspective, the core message of this article is that ‘an interactive interface is not necessarily a better interface.’ Many sites embed AI chatbots on their homepages, but in reality, search, filters, and navigation structures already work well. Adding a chatbot on top of these can actually increase interaction costs. The particularly interesting point is when the chatbot’s role is ‘supporting’ rather than ‘replacing.’ Chatbots excel in areas where traditional UI misses, such as complex product comparisons, context-dependent policy questions, and detailed product inquiries. Therefore, CIT interprets the adoption of AI features not as adding a UI, but as redesigning the service context.

Questions to Consider While Reading

• Q.What are the unresolved problems that a chatbot truly needs to address in our service, and which areas are sufficiently covered by existing search, filters, and navigation?
• Q.If we narrow the chatbot’s role to ‘contextual support’ rather than ‘full guidance,’ on which pages and at which moments would it be most appropriate to introduce it?
• Q.To prevent user misunderstanding of the chatbot’s capabilities, how should we design the initial message and example questions to effectively convey its information scent?