Apple's recent unveiling of a more capable Siri marks a significant milestone in AI development, achieved through strategic collaborations with Google and Nvidia. This new iteration of Siri leverages Google's Gemini foundation models, albeit without using the Gemini app, to enhance its contextual understanding and integration across Apple's ecosystem.
Siri's Architectural Evolution
The new Siri architecture is a complex system featuring both local and cloud-based models. Key components include AFM Core, AFM Cloud Pro, and ADM Cloud Images, which collectively represent a leap in quality and operational capability. Amar Subramanya, VP of AI at Apple, highlighted the innovative approach to handling parameters, which reduces memory strain and enhances efficiency.
Collaborative Model Development
Apple's collaboration with Google is evident in the co-developed models, where Apple's proprietary builds are integrated with Google's AI capabilities. This partnership allows Siri to process queries more efficiently, using a 'scarce model' approach to manage parameters without overloading device resources.
Leveraging Nvidia's Technology
To support the advanced models like AFM Cloud Pro, Apple has extended its Private Cloud Compute to include Nvidia's latest technology. This move ensures that Apple's stringent privacy and security standards are maintained while utilizing third-party cloud infrastructure.
"Apple devices can only talk to software signed by Apple," ensuring that Siri's operations remain secure and private.
The Future of Siri
This enhanced Siri is a testament to Apple's commitment to innovation through collaboration. By integrating cutting-edge AI technologies from Google and Nvidia, Apple aims to deliver a voice assistant that is not only more intelligent but also deeply integrated into its ecosystem, providing a seamless user experience.
