Following the keynote, Craig Federighi and John Giannandrea delved deeper into their AI initiatives in a discussion hosted by iJustine at the Steve Jobs Theater.
Privacy at the CoreFederighi and Giannandrea emphasized Apple's longstanding focus on user privacy, which they believe enables the company to utilize personal data in ways that other AI systems cannot. Apple Intelligence is designed to empower users rather than replace them, aiming for a more personalized and context-aware experience.
Federighi: "We want AI not to replace our users, but to empower them." Example: Apple Intelligence can provide personalized responses, such as travel times to a specific event, considering individual preferences and details. AI Integration and PerformanceApple’s approach involves integrating AI into specific, reliable use cases rather than broad, generalized AI applications. This method ensures a curated and effective user experience.
Giannandrea: AI features are supported on Apple's M-series chips and the A17 Pro in the latest iPhone models due to their advanced computational capabilities. Federighi: The A17 Pro’s Neural Engine is significantly more powerful than its predecessor, facilitating advanced AI functionalities. Ethical AI Model TrainingApple is transparent about the data used to train its AI models. They utilize a mix of public web data and licensed sources, ensuring ethical practices and paying for licensed content.
Giannandrea: Training data sets are created in-house by Apple, maintaining high ethical standards and user privacy. Private Cloud ComputingApple introduced Private Cloud Computing to handle tasks that require extensive computational resources while ensuring privacy.
Federighi: Only minimal, anonymized data is uploaded to servers that cannot log or store information. Apple ensures transparency by only allowing devices to communicate with servers running publicly published software, which third parties can verify. Integration with Third-Party AI ModelsApple is open to integrating third-party AI models, such as ChatGPT and Google's Gemini, to enhance Siri's capabilities. This allows users to access specialized data while adhering to the privacy policies of the respective AI providers.
Giannandrea: Developers can leverage app intents to add AI functionalities within apps, expanding the potential of Apple Intelligence. Future of Apple IntelligenceApple Intelligence is still in its early stages, with much more to come. The feature is expected to launch as a beta later in the year.
Federighi: "It's very early innings here... this is the beginning of a long and exciting journey." Giannandrea: The integration of AI is set to bring significant value and create exciting opportunities for both users and developers.Apple's AI efforts, as outlined by Federighi and Giannandrea, showcase a commitment to creating a powerful, privacy-focused, and ethically sound AI ecosystem.