After months of speculation, Apple Intelligence took center stage in June at WWDC 2024. The platform was announced following a flurry of news about generative AI from companies like Google and Open AI, raising concerns that the notoriously tight-lipped tech giant had missed out on the latest tech boom.
But contrary to such speculation, Apple has assembled a team to work on a typically Apple approach to artificial intelligence, and while the demo was flashy (Apple always loves a splash), Apple Intelligence is ultimately a very pragmatic approach to the field.
Apple Intelligence (yes, AI for short) is not a standalone feature. Rather, it is something that is integrated into existing products. It's really a branding effort, but the Large Language Model (LLM)-driven technology works behind the scenes. As far as consumers are concerned, the technology will primarily appear in the form of new features in existing apps.
More details will likely be revealed at the company's iPhone 16 event, which starts at 10 a.m. on Monday, when, in addition to the new iPhones, we're also expected to see updates to the Apple Watch, AirPods, and even new Macs.
What is Apple Intelligence?
Image credit: Apple
Cupertino's marketing executives have dubbed Apple Intelligence “AI for the masses.” The platform is designed to take what generative AI is already good at, such as generating text and images, and improve on existing capabilities. Like other platforms such as ChatGPT and Google Gemini, Apple Intelligence was trained on large-scale information models. These systems use deep learning to form connections between text, images, videos, music, and more.
The Text feature powered by LLM is offered as a writing tool. The feature is available in various Apple apps, including Mail, Messages, Pages, and Notifications. It can be used to provide summaries of longer pieces of text, proofread, and write messages with content and tone prompts.
Image generation is integrated in a similar way, though a little less seamlessly: users can instruct Apple Intelligence to generate custom emojis (Genmoji) in the Apple house style, while Image Playground is a standalone image generation app that uses your instructions to create visual content that can be used in Messages and Keynote or shared on social media.
Apple Intelligence also brings long-awaited improvements to Siri, the smart assistant that was around early but has been largely ignored for the past few years. Siri is now more deeply integrated into Apple's operating system. For example, instead of the familiar icon, you'll see a glowing light on the edge of your iPhone screen when Siri is working.
More importantly, the new Siri works across apps, which means, for example, you can ask Siri to edit a photo and insert it directly into a text message – a frictionless experience never before seen in an Assistant. On-screen awareness means Siri uses the context of the content you're currently working in to provide you with the right answer.
Who can get Apple Intelligence and when?
Image credit: Darrell Etherington
It's too early to tell how effective these features are — Apple's latest operating system was released in public beta this week, and Apple Intelligence isn't finished yet — but it's clear that Apple was pressured to promote the feature in June to allay concerns that the company had no AI-generating plans and to give developers a head start.
Though we saw a demo at WWDC, you'll have to wait until the fall to get a beta of Apple Intelligence, which coincidentally is also when the public versions of iOS/iPadOS 18 and Mac Sequoia will hit the App Store. It'll be free if you have any of the following hardware:
iPhone 15 Pro Max (A17 Pro) iPhone 15 Pro (A17 Pro) iPad Pro (M1 or later) iPad Air (M1 or later) MacBook Air (M1 or later) MacBook Pro (M1 or later) iMac (M1 or later) Mac mini (M1 or later) Mac Studio (M1 Max or later) Mac Pro (M2 Ultra)
It's worth noting that only the Pro version of the iPhone 15 will have access to it due to a flaw in the chipset on the standard model, but presumably the entire iPhone 16 series will be able to run Apple Intelligence once it's released.
Private Cloud Computing
Image credit: Apple
Apple takes a bespoke, small-model approach to training. Rather than relying on a one-size-fits-all approach used by platforms like GPT and Gemini, the company compiles datasets in-house for specific tasks, like composing an email. The biggest benefit of this approach is that many of these tasks are much less resource intensive and can be performed on-device.
However, this is not the case everywhere. More complex queries will make use of a new private cloud computing service. The company now operates remote servers that run on Apple Silicon, which it claims will provide the same level of privacy as consumer devices. Whether an action is performed locally or via the cloud is invisible to the user unless the device is offline, in which case remote queries will result in an error.
Apple Intelligence with Third-Party Apps
Image credit: Didem Mente/Anadolu Agency/Getty Images
Ahead of WWDC, there was talk of a planned partnership between Apple and OpenAI, but it turned out that the partnership was more about providing an alternative platform for things Apple Intelligence didn't really address, rather than strengthening Apple Intelligence itself — a tacit admission that there are limitations to building small model systems.
Apple Intelligence is free, as is access to ChatGPT, but users with paid ChatGPT accounts will have access to premium features not available to free users, which will likely be a big boost for the already thriving generative AI platform.
We do know that Apple is planning to partner with more generative AI services, and the company has revealed that the next on that list is Google Gemini.