Apple is set to enhance its AI capabilities by analyzing user data directly on devices, aiming to improve its Apple Intelligence platform while maintaining its commitment to user privacy. The initiative, reported by Bloomberg, marks a significant shift in Apple's approach to AI development.
Real data, better results: Traditionally, Apple has relied on synthetic data to train its AI models. But this method has limitations, as synthetic data may not accurately represent real-world user interactions, leading to less effective AI performance. To address this, Apple plans to utilize real user data from those who opt into its Device Analytics program.
Privacy considerations: The process involves comparing synthetic datasets to actual user data samples, such as recent emails or messages. The change in approach will allow Apple to refine its AI outputs, like email summaries, without direct access to personal information. By leveraging on-device processing and differential privacy techniques, Apple aims to enhance its AI capabilities while safeguarding user privacy.
Playing catch-up: The new system is being introduced in beta versions of iOS and iPadOS 18.5, and macOS 15.5. Apple's move comes as it seeks to catch up with AI rivals like OpenAI and Alphabet, which have fewer privacy restrictions. By analyzing data on customers' devices, Apple aims to improve its AI platform while ensuring user data remains on devices and isn't directly used to train AI models.