Generative UI is a powerful capability in which an AI model creates not just the content, but the entire user experience. Today we introduce a new implementation of Generative UI, which dynamically creates immersive visual experiences and interactive interfaces—such as web pages, games, tools, and apps—that are automatically designed and fully customized in response to any question, instruction, or prompt. These prompts can be as simple as a single word, or as long as detailed instructions are needed. These new types of interfaces differ markedly from the static, pre-defined interfaces in which AI models typically display content.
In our new paperGenerative UI: LLMs are efficient UI generators,We describe the basic principles that enabled the ,implementation of Generative UI and demonstrate the effective ,feasibility of this new paradigm. Our evaluations indicate that when ,generative speed is ignored, the interfaces from our ,generative UI applications are strongly preferred by human ,raters compared to standard LLM outputs. This work ,represents a first step towards fully AI generative user experiences, ,where users automatically get dynamic interfaces tailored to their ,needs, rather than having to choose from an existing ,application catalogue.
Our research on generative user interfaces, also referred to as generative interfaces, is now a reality in… Gemini application Through an experience called Dynamic Display and in AI mode Google search.
