It’s been a few months since WWDC 2024 and Apple’s announcement of its take on AI. As a long-time observer of Apple – I was a contributor and editor of Macworld Australia for many years – the thing I found most jarring was Apple’s very quick pivot.
WWDC 2023 was all about spatial computing. Apple announced the Vision Pro and told the world it was ushering in a new revolution in how we would interact with the world through digital devices.
“Spatial computing will enable devices to understand the world in ways they never have been able to do before. It is going to change human to computer interaction, and eventually every interface — whether it’s a car or a watch — will become spatial computing devices.”
Tim Cook, Apple CEO, 2023
Just a year later, events overtook Apple’s massive ambition. The computing revolution Apple anticipated and sought to lead petered out. OpenAI, with its many different generative AI tools, Google, Meta and others completely hijacked the revolution and put Apple into catch up mode.
For the first time in many years, Apple found itself as the challenger and not the leader. And despite its best efforts, Apple knew that it’s only way into the AI game was by partnering or leveraging someone else’s tech. This is what led the folks at 1 Infinite Loop to a partnership with OpenAI.
The OpenAI deal is a classic Apple move
Apple’s DNA, it could be argued, is to take inventions that already exist and apply them in ways that others haven’t thought of. Or to see the potential of something that others have missed.
This goes back to the first Mac in 1984, when Apple leveraged the Graphical User Interface and the mouse. These were inventions that came from the famed laboratories at Xerox PARC in Palo Alto.
The iPod, the device that helped transform Apple from a computer company into a global electronics and media giant, was not a new idea. There were many other portable music players on the market. But Apple saw a different way to bring the same core functionality to consumers. Paired with perhaps one of the best marketing campaigns ever conceived, it was a roaring success.
If you break down the iPhone, you’ll quickly discover that its constituent hardware and software were not new creations. The Gorilla Glass front was invented in the 1950s and produced by CorningWare. Multi-touch first appeared in the 1980s and touchscreen phones had been on the market for some time.
In each of these cases, Apple took things that were already available and put them together in novel ways. And that brings us to Apple Intelligence.
What makes Apple Intelligence special?
The integration of AI into Apple’s core operating platforms is a classic example of Apple taking something that is widely available and applying its own touches to differentiate it from other products in the market.
At least that’s what Apple is hoping.
(For the purpose of this article, what I say AI I mean Artificial Intelligence. When I mean Apple Intelligence I’ll be explicit.)
When Apple Intelligence is asked a question, it will first try to answer by using AI models that are on the device. Eventually (Apple’s timeline for when various AI features will be shipped to users seems quite nebulous) we’ll see that capability on iPhones (you’ll need an iPhone 15 Pro or later for that), the iPad or a Mac (assuming you have a device with an M-series processor).
The on-device models will learn from how you specifically use your device and the data that’s stored on your device. It will also provide a number of ChatGPT-like tools such as the ability to summarise and rewrite content. And it will prioritise messages, summarise long email chains and provide lots of other handy features.
I suspect that once we start using Apple Intelligence, it will become indispensable.
If the on-device model can’t answer the question, the query will be sent, with full end-to-end encryption to Apple’s servers, dubbed Private Cloud Compute, where it is processed. This will, presumably, leverage OpenAI models that Apple will license – although Apple’s documentation about this is not explicit – as well as its own.
If Apple’s servers can’t handle the query, it will then be securely passed to OpenAI’s servers. This is done in a way that makes it impossible for the initial query to specifically identified as coming from a specific person or device.
So, what makes Apple Intelligence special?
Apple is taking a cautious approach to how AI is being deployed through iOS, iPadOS and macOS. There is a strong focus on privacy and security with Apple leveraging exisiting security technologies such as Secure Enclave as well as robust encryption and on-device processing wherever possible.
Unlike online tools such as ChatGPT, based on what Apple has said thus far, you are giving very little over to third parties.
Privacy and security seem to be the secret sauce Apple hopes will help it enter the AI game.
Parallels to Apple’s OpenAI deal
For many years, Apple has licensed technology from Qualcomm to provide models for its devices. At the same time, Apple has been developing its own modem chips – thus far with limited success.
Apple announced that it was shifting from Intel processors to its own silicon in 2020 and has transitioned all its products. The development of Apple Silicon happened in parallel with the Intel partnership.
This feels a lot like that to me. Apple has partnered with an acknowledged leader in a field while continue to develop its own technology. Apple’s investment in Private Cloud Compute is built on Apple Silicon. And that tells us Apple has been thinking about AI for some time.
It may take a few years to get it right (remember the introduction of Apple Maps?) but Apple will, in time, wean itself from OpenAI and transition to its own AI platforms. Apple’s former marketing lead and the current lead for the App Store, Phil Schiller, has been appointed as an observer on the OpenAI board. While this doesn’t give Apple a say in how OpenAI operates, it does give them valuable intelligence into what OpenAI is planning.
Apple won’t have its own competitive AI service ready in a year. It probably won’t have caught up to OpenAI, Meta and Google in three years. But, eventually, it will become less reliant on a partner, just as it did with Intel and Google. And that’s when Apple Intelligence will truly come of age.
Anthony is the founder of Australian Apple News. He is a long-time Apple user and former editor of Australian Macworld. He has contributed to many technology magazines and newspapers as well as appearing regularly on radio and occasionally on TV.