With Apple Intelligence, Apple introduced AI functions on iPhones, iPads and Macs last year. This allows users to access writing assistants or have messages summarized.
The iOS 18.4 update added, among other things, an option, so-called. It is an emoji generated by AI. So far, many of the functions seem behind the.
That happened:To change this, Apple wants to recognize trends better. In one article hasApple insightplaced in the underlying technology.
In any case, the consent of the users is necessary for the collection of data.
Balance between the privacy of users and the use of data
AI needs access to the large amount of data as large as possible. Apple indicates that you do not use personal data or activities of users for training.
Nevertheless, data from users play an important role for Apple, including in the detection of trends. Apple relies on the so -called "differential privacy", in German differential data protection.
Good to know: Differential data protection uses data so that the query does not allow any conclusions to be drawn about certain people.
Apple is already using the technology to improve gene mogji and the creation of texts.
Improvement of genemoji
For Apple, the detection of popular entries for creating genemoji is important to recognize trends.
In summary, the detection works as follows:
- Apple sends inquiries to devices to clarify whether a certain input is popular.
- Devices answer anonymously with noise. Rowing means that a device does not always with
Yes, is known
orNo, unknown
replies, but also withsomething random
.
If there is a positive feedback from many devices, the input can be displayed when using the AI tool in order to react to trends.
Apple points out that identifying information, such as the IP address of the device, is removed.
Improving the creation of text
Apple uses a modified system for the detection of trends when creating texts.
Apple is artificially created for this application, whichis representative of merged trends in real user data
, but without the fact that actual user data is queried.
A representative part of a generated text is then sent to devices by Apple users to find out how popular it is.
The most popular texts can then be used to train AI or for further refinement.
Now your opinion is asked. Do you already use AI tools like Apple Intelligence on your devices? Or do you have concerns about privacy? Please write us a comment.
