Apple plans to implement a customized version of the Gemini model on its Private Cloud Compute servers for secure analysis, according to The Information. The company has previously indicated that the upgraded Siri will possess significantly enhanced capabilities. For instance, Siri will be able to respond to personal inquiries, such as ‘Find the book recommendation from Mom,’ without processing data in the cloud. Instead, it will intelligently access locally stored information to provide quick answers. The integration of Google Gemini into Siri is reportedly structured into three components: the query planner, knowledge search system, and summarizer. These components will utilize Gemini models to improve Siri’s understanding of user requests, enabling it to deliver more natural and context-aware responses.
Apple will maintain the use of its Foundation Models to protect personal information on devices. The privacy-centric company has assured that Google’s models will remain on its Private Cloud Compute servers and will only operate on Apple infrastructure. Siri’s enhancements may also draw from the search system associated with the Gemini model. This will allow Siri to better handle general knowledge inquiries, providing richer and more contextually relevant answers regarding current events, trivia, and facts. However, the integration of Gemini does not imply that Apple will rely on ChatGPT or third-party applications for Siri, nor will it provide basic, unrefined responses. Siri’s answers will continue to originate from within the Apple ecosystem, which Apple plans to emphasize in its marketing.
Although the backend will leverage Gemini technology, Bloomberg reports that Apple does not intend to disclose this partnership, aiming to present Siri’s improvements as solely its own advancements. The incorporation of Google’s Gemini technology will fortify Siri’s backend systems in areas where Apple’s own large language models have previously struggled.