Apple’s integration of Google’s Gemini AI models into Siri has sparked questions about the nature of this collaboration. While Siri now leverages Gemini’s capabilities, it remains distinct from Google’s own Gemini Assistant.
Google’s terminology can be confusing, as ‘Gemini’ refers both to a series of AI models and to its intelligent assistant. However, Siri is not a rebranded version of Gemini Assistant; it utilizes Gemini models to enhance its functionality while maintaining its unique identity.
Apple has developed its own set of models, known as Apple Foundation Models (AFM), in collaboration with Google. These include four custom versions of Gemini optimized for Apple Silicon and a fifth, more advanced model, which is essentially Google’s standard model running on Google servers but likely trained with different data sets.
Privacy remains a cornerstone of this integration. Two of the AFM models operate entirely on-device, ensuring user data never leaves the device. The other two run on Apple’s Private Cloud Compute (PCC) servers, designed so that no data is retained or accessible by Apple or Google. The most powerful model runs on dedicated Google servers but adheres to the same PCC architecture, maintaining stringent privacy standards.
This collaboration signifies a strategic move by Apple to enhance Siri’s capabilities without compromising user privacy. By customizing Gemini models and implementing robust privacy measures, Apple aims to deliver a more intelligent and secure user experience.
Source: 9to5Mac
