Apple Rebuilds Siri with LLM AI Overhaul Amid Internal Setbacks

Apple is undertaking a major overhaul of its Siri voice assistant by rebuilding it entirely on a large language model (LLM) architecture to better compete with AI rivals like Google and OpenAI. This shift comes after the initial Apple Intelligence features faced significant delays, internal resistance, and technical challenges, stemming from Apple's late entry into generative AI and hesitance to invest heavily without guaranteed outcomes. Key issues included difficulties merging new AI capabilities with Siri's aging infrastructure, reluctance from senior leadership to commit to AI investments, and Apple AI chief John Giannandrea's skepticism about consumer demand for chatbot features. The new project, internally called 'LLM Siri' and developed by Apple's Zurich AI team, aims to create a more conversational and information-synthesizing assistant that can also access and integrate web data from multiple sources. Giannandrea has reportedly been removed from direct involvement with Siri and other AI projects, with Apple executives contemplating his retirement. Overall, Apple is focusing on a ground-up redesign leveraging LLMs and enhanced data synthesis to revitalize Siri after previous strategic missteps and development setbacks.