Apple Intelligence: A Cautious but Potentially Transformative Approach to AI Integration

Despite unforced errors, the future of Apple Intelligence could be bright

Ask a few random people about Apple Intelligence and you'll probably get quite different responses. One might be excited about the new features. Another could opine that no one asked for this and the company is throwing away its reputation with creatives and artists to chase a fad. Another still might tell you that regardless of the potential value, Apple is simply too late to the game to make a mark. The release of Apple's first Apple Intelligence-branded AI tools in iOS 18.1 last week makes all those perspectives understandable. The first wave of features in Apple's delayed release shows promise -- and some of them may be genuinely useful, especially with further refinement. At the same time, Apple's approach seems rushed, as if the company is cutting some corners to catch up where some perceive it has fallen behind. That impatient, unusually undisciplined approach to the rollout could undermine the value proposition of AI tools for many users. Nonetheless, Apple's strategy might just work out in the long run. I'm basing those conclusions on about a week spent with both the public release of iOS 18.1 and the developer beta of iOS 18.2. Between them, the majority of features announced back in June under the "Apple Intelligence" banner are present. Let's start with a quick rundown of which Apple Intelligence features are in each release.

Apple Intelligence is the most staggered launch in Apple history -- is that a good thing?

Apple's AI strategy is purposely slow and deliberate, but it could backfire Everywhere you look these days there's the glowing animation of Siri and the name Apple Intelligence. It's dominating all the ads for the iPhone 16 series so far, whether it's Patrick Mahomes hurdling other customers to grab his new iPhone or a couple in bed fawning over the features while creeping out their kids. Yup, Apple seems to be pinning all of its would-be upgraders' hopes on Apple Intelligence, which is the company's umbrella title for all of its AI features. Current offerings range from Writing Tools and asking Siri iPhone questions to creating Memory Movies. Plus, Apple Intelligence is available on all Macs and iPads with an M1 chip or newer. But there's still a lot of Apple Intelligence features not available yet -- Visual Intelligence, ChatGPT integration, Genmoji and Image Playground are coming in iOS 18.2 in December. And other features are promised in future iOS 18 updates. Meanwhile, Bloomberg's Mark German recently reported that some Apple employees believe that the company is around two years behind in artificial intelligence development. So is Apple Intelligence on the right track or in trouble? I can definitely understand why customers might expect all Apple Intelligence features to work from day one, and it's clear that the likes of OpenAI, Google and Anthropic have more advanced models at this stage. "Apple's AI models are demonstrably behind the competition, and it is just now adding imaging editing features that Samsung and Google introduced two generations ago," says Avi Greengart, founder and lead analyst at Techsponential. "However, I'm not convinced that this means that Apple is behind in AI overall." Greengart told me that having a lead in AI technology is only meaningful if it leads both to consumer benefits and a sustainable business model. Apple's monetization model is selling premium hardware and software/ecosystem subscriptions, so for now Apple Intelligence is about getting people to upgrade to iPhones and other devices that support the platform. Any possible subscriptions would come later once Apple gets a sense for what users want. I asked Bob Borchers, Apple's vice president of worldwide marketing, about the perceived gap between Apple and the competition on AI. "This is very much the early innings of generative AI," he told me. For better or worse, Apple's approach to AI is what truly sets it apart from the competition. It's not a separate thing that's tacked on; instead, it's deeply integrated into the apps and tools you use everyday. For example, Apple Intelligence will boil down your notifications into a quick summary to help you prioritize. Or if you want to remove an unwanted person or object from your photo, you'll see the Clean Up icon there along the bottom of the screen. "What we've done with Apple Intelligence is really look at how can we help our users do the things that they care about most, simply and easily, and really to make that as integrated and local to the experience as possible," says Borchers. "And to do that, we started by building our own large language and diffusion models here at Apple, and then we specialized them for everyday tasks." Still, Apple's cautious approach to AI image generation in particular seems to put them at a disadvantage when compared tools you'll find in the latest Samsung Galaxy S24 and Pixel phones. For example, with the Pixel 9 series, you can completely reimagine what's in the frame with a text prompt. But Apple seems to be deliberately holding back some features or simply deciding not to include them in order to avoid potentially harmful or misleading images from circulating. In an interview with The Wall Street Journal, Apple software chief Craig Federighi said that "It's important to us that we help purvey accurate information, not fantasy." And that brings us to one of Apple's four key Responsible AI Principles, which focuses on taking precautions during the design, model training and feature development process to identify how Apple's AI tools may be misused or lead to potential harm. The other three principles relate to empowering users with intelligent tools, representing Apple's users by building personal products that avoid perpetuating stereotypes and protecting privacy. "Our view has always been that we bring products to market when they are ready, and when we feel like we can do them in a way that is consistent with our values," said Borchers. "That means making generative AI available in a useful way, in a responsible way, in a way that's private." Private Cloud Compute is a big part of Apple's AI strategy, which handles requests that go off-device to interact with larger models, including ChatGPT. Apple is so confident that personal user data can't be accessed by anyone that it recently offered up to $1 million to anyone who could compromise its security. So is Apple making the right call with its measured approach to Apple Intelligence or is it already being passed by? For example, it's hard not to be impressed with ChatGPT Advanced voice and the ability to chat with what seems like a friend, complete with your choice of intonation on the fly. For Neil Cybart, an AI analyst with Above Avalon, it's important to keep in mind that Apple isn't targeting early adopters but the masses. "A gradual rollout for Apple Intelligence is the right move given how we will likely see roughness around the edges as people begin to kick the tires on features. The average Apple user probably wasn't going to suddenly embrace everything Apple Intelligence has to offer anyways," says Cybart. "An Apple Intelligence launch that is spread out over the next year doesn't concern me." As Cybart notes, not only is Apple Intelligence a work in progress on the iPhone, iPad and Mac, Apple hasn't started extending Apple Intelligence to the rest of its ecosystem yet, whether it's the Apple Watch with fitness and health, the AirPods or the Vision Pro. So where does that leave Apple users now? According to CEO Tim Cook during Apple's most recent earnings call, "users are adopting iOS 18.1 at twice the rate that they adopted 17.1 in the year-ago quarter." Granted, that's based on only three days' worth of data, but it shows that the glow around Apple Intelligence seems to be spurring excitement.

Apple Intelligence will help AI become as commonplace as word processing

When Apple's version of AI, branded as Apple Intelligence, rolls out in October to folks with the company's latest hardware, the response is likely to be a mix of delight and disappointment. The AI capabilities on their way to Apple's walled-garden will bring helpful new features, such as textual summaries in email, Messages and Safari; image creation; and a more context-aware version of Siri. But as Apple Intelligence's beta testing has already made clear, the power of these features falls well below what is on offer from major players like OpenAI, Google, and Meta. Apple AI won't come close to the quality of document summary, image or audio generation easily accessed from any of the frontier models. But Apple Intelligence will do something none of the flagship offerings can do: change perceptions of AI and its role in ordinary life for a large portion of users around the world. The real impact of Apple AI won't be practical but moral. It will normalize AI, make it seem less foreign or complex. It will de-associate AI from the idea of cheating or cutting corners. It will help a critical mass of users cross a threshold of doubt or mystification about AI to forge a level of comfort and acceptance of it, even a degree of reliance. Generative AI has faced two problems since ChatGPT was unveiled in 2022. Many have wondered what it's really for or whether it's truly useful, given hallucinations and other issues that are rooted in training data. Others have doubted the ethics of using AI, seeing it as a form of cheating or copyright infringement. But as we have learned in recent months, language models are most effective when they work on our own documents and data, as with platforms like NotebookLM or GPT4o, which can now handle upwards of 50 to 100 books' worth of material we upload. The output of the prompts we run -- in the form of article or lecture summaries, reports, slide decks and even podcasts -- is much more accurate and useful than what came out of earlier chatbots. Apple Intelligence capitalizes on this insight by pointing most of its AI functionality at user data, rather than data on the web. With Apple Intelligence working mainly on our own data, much of its output will likely mirror the higher quality of output we're seeing with tools like NotebookLM -- compared to AI that works mainly on large bodies of anonymous training data, like ChatGPT in its early days. Having AI work mostly on user data -- and doing it frequently -- will forge a new association in people's minds between generative AI and personal information, rather than miscellaneous training data. It will likely cause us to see AI as something integral to our personal routines, like reading email or the morning news. This, in turn, will make using more powerful tools like GPT4o or Claude more socially and ethically acceptable. Once we're in the habit of using AI to summarize or edit our email, condense articles on the web into pithy summaries or edit images in Photos, we'll think less about the propriety of using NotebookLM to prepare a first draft of a memo or report, or using Dall-E to create images. Apple has a long history of making complex technologies more accessible to everyday users, and that is their goal for AI. When word processors first appeared in the late 1970s and early 1980s, there was similar uncertainty about the propriety of using them to help us write things -- a belief that something authentic or human about writing by hand would be lost. For many, computers themselves were too daunting to embrace. But Apple's Macintosh personal computer helped domesticate and normalize using computers to write with its graphic user interface and WYSIWYG feature ("what you see is what you get"). Eventually, writing would become so closely associated with word processing that we find it hard to imagine the one without the other. Apple Intelligence could do for generative AI what the Mac or graphic user interface did for personal computers: help tame it, and make it seem ordinary and acceptable. Apple's marketing team hints at this in their tagline for Apple Intelligence, "AI for the rest of us." If history is any guide, Apple will play a key role in changing how we think about AI. Doing many of our basic tasks without it may soon seem unthinkable.