Apple and Google have forged a multi-year partnership, with Apple leveraging Google's Gemini models as the cornerstone for its own AI models. This strategic alliance will significantly impact Apple's AI capabilities, particularly Siri, which is set to receive a long-awaited AI upgrade later this year. The joint statement, available at [link], reveals that Apple has chosen Google's AI technology as the optimal foundation for its Foundation Models, also known as Apple Intelligence. This decision is based on extensive research, ensuring a robust and efficient AI foundation for Apple's products.
The end-user experience is unlikely to change immediately. Gemini, a closed-source model typically running on Google Cloud, will be seamlessly integrated into Apple's ecosystem. Apple may utilize its own servers to run the AI model, providing a personalized experience without the user's awareness of the underlying technology. The upgrade promises to enhance Siri's capabilities, enabling it to work with a larger language model, resulting in faster task completion and more tailored output.
The exact implementation details remain undisclosed. Apple might employ a process called 'distillation,' where a compact model learns from the complex answers of a large model, potentially using Gemini to train its own smaller models. However, the statement's reference to Gemini as the 'foundation for foundation models' could be a marketing strategy, allowing Apple to maintain control over its AI development while leveraging Google's technology.
Financial implications are significant, though details are scarce. Bloomberg has speculated that Apple might pay Google a substantial annual fee, possibly in the billions, for access to the 1.2 trillion-parameter Gemini model. Despite their competitive nature, Apple and Google's collaboration extends beyond this deal. For instance, Apple's Safari browser has featured the Google search bar for years, showcasing their mutual reliance and potential for further joint ventures.
