Breaking Boundaries: Apple and Google Team Up for Revolutionary Siri AI in iOS 27

Following the WWDC keynote, Apple’s Siri team, under the leadership of Craig Federighi, engaged in a tech talk to delve into iOS 27 and the enhanced Siri AI.

During the discussion, Federighi provided insights into Apple’s collaboration with Google.

Accompanying Federighi were Amar Subramanya (VP of AI), Mike Rockwell (Siri lead), and Sebastien Marineau-Mes (software VP).

Regarding the Google collaboration, Federighi clarified:

Apple does not incorporate the Gemini app or client code from Google in our operations. We do not utilize Google’s models or infrastructure, including the knowledge base from Google Search. Our usage of the Google Assistant is zero.

Our system is built on the integration of the Assistant experience into iOS, iPadOS, and macOS, as demonstrated in the keynote. The Siri app acts as a bridge to continue or initiate conversations, leveraging on-device models and Private Cloud Compute for advanced queries.

The privacy-focused architecture includes the System Orchestrator, App Toolbox, Spotlight Semantic Index, and on-device models for speech understanding, voice synthesis, and contextual awareness.

Apple’s World Knowledge Service provides accurate responses to queries involving current events or general knowledge.

Subramanya further elaborated:

The 3rd generation of Apple Foundation Models (AFM) in collaboration with Google introduces a range of on-device and server models, optimized for quality and performance. On-device models like AFM Core and Core Advanced enhance user experiences, while server models like AFM Cloud and Cloud Image cater to specific tasks like image editing.

The AFM Cloud Pro model, developed with Google and Nvidia, handles complex tasks with privacy guarantees. The goal is to match user requests with models for optimal responses.

Stay tuned for more updates…

Chance’s favorites:

Follow Chance: Threads, Bluesky, Instagram, and Mastodon.