Apple has gained access to Google’s Gemini system, allowing it to create smaller, task-specific AI models that operate on iPhones and iPads without internet dependency. This distillation access enhances Apple’s privacy stance, as users’ data remains on-device rather than sent to Google’s servers. Major changes to Siri, including a chatbot codenamed Campos, are expected to be announced at WWDC in June, although distilled models have yet to be deployed; Siri currently relies on cloud-based responses. The knowledge distillation process enables Apple to build efficient models that replicate Gemini’s capabilities while prioritizing user privacy. This partnership has the potential to improve Siri’s performance and close the gap with competitors using advanced on-device AI solutions. Apple’s Foundation Models team continues its in-house AI development alongside the Google collaboration, aiming for meaningful enhancements by WWDC 2026. The upcoming conference will be critical to assess the effectiveness of this strategic alliance.
Source link
