The next version of iOS 18 will be the AI ​​version, this is an open secret. The first hints of this update tell us about an evolution of Siri that will eventually see Apple’s voice assistant reach a level similar to Google’s. It’s a particularly difficult one as the assistant turns into Gemini (even though it’s still pretty green for now).

The question was whether Apple wanted to create a model for connecting to servers, with the security implications this entails, or whether it would implement learning models entirely natively. According to Gurman, Apple’s artificial intelligence will not leave the device.

iOS 18, AI version. Apple has been sleeping on artificial intelligence for years. Meanwhile, manufacturers are betting on some of their own functions, such as collaborations with Samsung, Google, as in the case of Nothing, or collaborations with OpenAI and ChatGPT.

Apple is afraid of missing this train, and we know that it is in talks with giants such as Google and OpenAI in order not to fall further behind. These plans point to the future, but in the iOS 18 version we will see this year, Apple’s first leap in artificial intelligence will already take place. A system that allows the system to act as support to the user and is not an implementation of more complex generative AI.

Privacy requires local execution. According to Gurman, Apple’s AI functions will run in a “massive offline model” that will be run from the device itself. This is an important move if Apple wants to differentiate itself from its competitors: not requiring an internet connection and not sending any data from the device.

Advantages and disadvantages. Beyond the extra privacy, there’s a big advantage to running native models that Apple is always worried about: being faster than others. One of the main problems AI currently has is with loading times. The device makes a request to the server and returns a response, which in some cases takes several seconds.

By running the model locally the response will be practically instantaneous. Likewise, by working without an internet connection, artificial intelligence can be used in areas where we do not have network coverage or WiFi connection. On the other hand, a 100 percent native model will always be less capable than a cloud-connected model. If there is no internet connection, it will not be possible to know some information in real time.

No word on what you can do. While we wait for July 10, the date when WWDC kicks off for Apple developers, where we hope to learn about this AI-powered version of iOS 18, we still don’t know what the system can do.

Everything seems to indicate that Siri can send and suggest automatic replies, simple rendering tasks (images, voice enhancement, text?), and improvements in both response times and quality.

Image | Xataka

in Xataka | Gemini Code Assist: what is it, how does it work, and what can this Google tool for developers do?