Apple details how it plans to improve its AI models by privately analyzing user data

After criticism of EU products, especially in areas warning summaryApple on Monday Detailed With the help of synthetic data, it is trying to improve the AI ​​models by analyzing the user information.

To use an approach called “Differential privacy“The company will initially set the devices of synthetic information and then requests to share with device analysts with device analysts), how accurate its models are, and then improve them.

“Synthetic data was created to imitate the format and important features of user data, but there are no users created content,” the company was written in the blog post. “We start creating great synthetic messages on various topics to treat a synthetic email set of a representative […] Then, we call a structure of every synthetic message that captures some basic dimensions of the message such as language, subject and length. ”

The company said that these announcements are sent to a small number of user devices that match the device analytics and compare them with the e-mail sample to say the devices, then the most accurate of the appedings.

The company uses this approach to improve this approach to improve this approach and the future image playground, image bar, memories and writing tools, as well as synthetic information for visual intelligence. Apple said that the users who choose to share device analysts with synthetic data to improve email summaries will also question.