From GUI to CUI: Exploring the Optimal Human-Machine Interaction Paradigm in the Age of AI
Most AI-powered applications have adopted the Conversational User Interface (CUI) for interaction, making it seem as though human-AI interaction is synonymous with conversation. We’ve even heard some SaaS vendors claim they will switch their products entirely to CUI. So, what advantages does CUI offer over the traditional Graphical User Interface (GUI)?
The primary advantage of CUI is its **low learning curve**. Users do not need to learn all the functionalities displayed on the screen (as in GUIs). Instead, they only need to express their needs clearly through language.
CUI is also more **flexible**. GUIs can respond to fixed user services, and the development cost for long-tail custom requirements is high. Typically, GUIs serve as a proxy for API service calls and returns, with this proxy being static. On the other hand, large language models (LLMs) can call tools (APIs) flexibly (though the current accuracy of these calls still leaves much to be desired).
GUIs display information in a structured manner (layout and formatting), multimodally (visual charts, images, videos), which is more human-friendly. For example, most people prefer watching movies to reading books.
