Gurman: Apple Intelligence Feature Rollout to Stretch into 2025

I feel like this is fairly obvious, but Bloomberg’s Mark Gurman reports that most of the AI features Apple recently announced will ship over several months rather than arriving all at once in the iPhone, iPad, and Mac OS udpates it will ship in September.

This rolling release of Apple Intelligence, which I think of Slow Cooked in honor of Apple CEO Tim Cook, stands in sharp contrast to the drunken and chaotic way in which Microsoft planned to deliver Recall and its other AI advances in its own platform. And is entirely in keeping with the Cook-led Apple does things.

Apple delivered the first beta releases of iOS 18, iPadOS 18, and macOS Sequoia to developers last week–and seriously, doesn’t WWDC suddenly feel like it was a month ago–but these releases don’t provide any of the Apple Intelligence advances it showed off at the time. Those won’t start arriving in pre-release form until the second beta, Gurman says. That’s the version that’s given out to consumers who wish to opt in as well.

But when these platforms arrive in September, the Apple Intelligence features will still be in preview–just as Microsoft planned for Recall on Copilot+ PCs–and they will work with only a subset of existing devices, and only in U.S. English. Some features will actually require a waitlist, Gurman says.

Apple didn’t announce this slow rollout during its WWDC keynote last week, and it has deployed certain individual features over time in the past. But the Apple Intelligence rollout is nonetheless unique because of its vast range of features over its three most important platforms. And while some features will be available in late 2024, some won’t appear until 2025, Gurman claims.

The delay will give Apple time to catch up: Its AI models still need to be trained more exhaustively, and Apple is still building out the Apple Silicon-based cloud infrastructure it bragged about at the show. The firm is also still negotiating with other AI partners, like Google, so it can add the Gemini models to the growing stable of on-device and cloud-based AI models it will provide to customers.

Many of the new Siri capabilities that Apple promoted at WWDC won’t arrive until 2025, Gurman says. These include the functionality that most closely resembles Microsoft’s Recall: The semantic indexing that will help Siri understand information on your devices and then the related finding of information on devices based on context. But also other features for precise control of apps (finding a photo of a certain friend in a red jacket, for example), meeting summarizations, and on-screen awareness, by which Siri can take actions using what’s on your device display as context. (Gurman uses the example of someone looking at a photo of a basketball player and asking Siri, “How many points did he score last night?”)

Apple Intelligence support for non-English languages won’t arrive until 2025, while some features, like ChatGPT integration, new Mail features, and Swift Assist for developers (Apple’s GitHub Copilot clone) will arrive later in 2024, Gurman says. Some non-AI features will also be delayed past the initial iOS 18, iPadOS 18, and macOS Sequoia releases, too, like spatial video editing, support for five simultaneous sports streams in Apple TV, and some new Home features.

Gurman also reports that Apple will follow up its success making the new iPad Pro thinner with thinner new versions of the MacBook Pro, Apple Watch, and the iPhone 17, all in 2025 or later.

“The plan is for the latest iPad Pro to be the beginning of a new class of Apple devices that should be the thinnest and lightest products in their categories across the whole tech industry,” he writes.

He also confirms his earlier report that Apple is not paying OpenAI for its use of ChatGPT but will instead reward the firm with exposure to its huge and loyal customer base. The hope is that many will sign up for a $20-per-month ChatGPT Plus subscription.