Apple develops device-side large-scale language model to improve AI response and privacy protection

News on April 22, according to the latest report from Bloomberg reporter Mark Gurman, Apple is currently working on developing a large language model (LLM) that runs on the device. The goal of this innovative initiative is to increase the responsiveness and enhance user privacy of upcoming generative AI features.

In Gurman's "Power On" newsletter, he specifically mentioned that this LLM under development will become a core component of Apple's future artificial intelligence capabilities. In stark contrast to current mainstream cloud-based AI services, this new model is designed to run entirely on the user's device. This design concept has significant advantages in providing faster service response times and superior privacy protection.

#However, because the model runs on the device, Apple’s AI tools may not perform as well in some aspects as those running directly in the cloud. Nonetheless, Gurman has reported that Apple may be able to make up for this potential performance gap by partnering with other AI service providers such as Google to bring in its advanced technology. Just last month, Gurman reported that Apple was in discussions with Google about integrating its Gemini AI engine into iPhones running iOS 18.

According to the cloud-based solution, the data editor learned that the biggest highlight of device-side processing is that it can provide faster response speed and better privacy protection. In addition, Apple's artificial intelligence technology marketing strategy is expected to focus more on how to actually help users improve the efficiency of their daily lives, rather than just emphasizing the responsiveness of the technology itself. More details on Apple’s artificial intelligence strategy and major software update plans are expected to be announced at the upcoming WWDC conference in June.

The above is the detailed content of Apple develops device-side large-scale language model to improve AI response and privacy protection. For more information, please follow other related articles on the PHP Chinese website!