Apple is set to improve its email summarization AI by harnessing synthetic data—computer-generated emails—to maintain customer privacy. This approach allows Apple to contrast its methods with the practices of Meta, which is resuming AI training on actual user posts unless they actively opt out. By using a large language model, Apple will create synthetic email variations that mimic real data format, enabling enhanced model training while ensuring user data security through techniques like differential privacy.
Synthetic data are created to mimic the format and important properties of user data, but do not contain any actual user generated content.
When creating synthetic data, our goal is to produce synthetic sentences or emails that are similar enough in topic or style to the real thing to help improve our models for summarization, but without Apple collecting emails from the device.
Collection
[
|
...
]