Google Cloud has introduced a powerful new feature—Grounding with Google Search—available in both Google AI Studio and the Gemini API. This feature enhances the Gemini generative AI models by incorporating live data from Google Search, ensuring that the AI delivers more accurate and up-to-date responses. The new capability is designed to make AI outputs not only more reliable but also directly linked to the sources that support the information, offering users added confidence in the model’s responses.
With Grounding, users now receive not just a response from the AI model, but also direct links to the sources that ground those responses. Additionally, search suggestions are provided that guide users to related search results, ensuring they have access to further context if needed. This feature, which is available for all generally available versions of Gemini 1.5, gives AI applications the ability to integrate real-time, factual information into their outputs, making them more dynamic and applicable across a broader range of use cases.
For developers looking to experiment with this new feature, Google AI Studio allows for free testing of Grounding with Google Search under the “Tools” section. For those using the Gemini API, the grounding feature can be accessed by enabling the google_search_retrieval tool. There’s a cost for the API access, with a paid tier priced at $35 per 1,000 grounded queries, offering developers a powerful tool for building more robust AI applications.
The benefits of grounding are significant. Not only does it improve the accuracy of the AI’s responses by pulling from the latest, factual information available on the web, but it also increases transparency by providing supporting links. Grounding enables AI models to handle a wide variety of scenarios, bringing a richer and more contextually relevant answer to users’ questions, all while enhancing trust through verifiable sources. This advancement in real-time information retrieval opens new possibilities for developers and businesses leveraging AI technologies.