Apple says its AI is different, but is it really?

Apple Intelligence, Apple’s pitch for artificial intelligence (AI) available directly on your iPhone, iPad, or Mac, is coming later this fall and is currently available as part of the iOS 18, iPadOS 18, and macOS Sequoia betas. Apple Intelligence is marking an exciting step for Apple as it competes with the impressive AI features Google continues to add to its Pixel devices.

After spending the summer experimenting with Apple Intelligence, hearing the thoughts of Apple commentators from various technology publications, and getting a recap during the company’s September event, I felt now is a good time to explore where Apple’s AI efforts stand, how the company’s approach to AI is unique, and where its future AI ambitions should go.

The state of AI on Apple’s platforms

Consumers generally don’t like it

Overwhelmingly, I’m tired of being pitched that everything is better with AI integrated into it. Like the crypto and blockchain craze last year, AI is currently used more as a buzzword to boost share prices than any valuable technology. At least as it stands today, AI is generally just generative AI, or the ability to create text, images, and videos from a prompt. It’s impressive technology, absolutely, but under most circumstances, generative AI only takes away from the creation that many of us enjoy rather than completing the chores and tasks we don’t.

That’s not to say there isn’t value provided by these tools. I’ve found countless efficiency benefits from tools like Descript, ElevenLabs, ChatGPT, and Perplexity. At the same time, some apps like my email client Spark are trying really hard for me to subscribe to a paid plan to unlock AI features when, in reality, I’d be more likely to pay to remove what I consider utterly bothersome AI ads. In fact, it seems that consumers mostly agree with me, as around 85% of people are actually turned off by AI-branded products.

It’s safe to say that plenty of AI tools are available on Apple’s operating systems from countless apps and services without the company doing anything. So, does Apple Intelligence even have a spot to compete on the company’s own operating systems? Yeah, I think so.

Apple’s approach to AI

The focus is on “personal context” and privacy

On the surface, Apple Intelligence seems similar to other generative AI tools. For example, Apple Intelligence’s Writing Tools allow users to rewrite content into a more friendly, professional, or concise voice. Additionally, Writing Tools offer the ability to proofread content similar to Grammarly but with less granular settings on tone. Writing Tools can also summarize content, along with reformatting it into key points, tables, and lists.

In addition to Writing Tools, Apple Intelligence brings image generation to your devices through three new tools: Image Wand, Image Playgrounds, and Genmoji. Image Wand allows users to create more polished images from sketches made in the Apple Notes app, while Image Playgrounds lets you create new images in the styles of sketch, animation, or illustration based on a prompt. You can even use Image Playgrounds to create cartoon-style pictures of contacts. With Genmoji, you can generate your own emojis by typing a prompt like “cowboy cucumber,” presumably creating a cucumber with a cowboy hat or something of the sort.

Apple Intelligence is also improving Mail, Safari, notifications, Photos, Notes, and Siri. Apple Mail, Safari, and notifications can generate summaries of their respective content and surface priority information. Notifications can also be filtered and prioritized thanks to a new ‘Reduce Interruptions’ focus mode powered by Apple Intelligence, which filters up time-sensitive and important notifications.

Apple

Messages and Mail are getting ‘Smart Replies’ to make replying to conversations quicker and easier. Apple Photos users can search for photos and videos using natural language and generate custom movie memories of people in their photo libraries based on a prompt. At the same time, audio recordings in Notes get transcriptions along with being the app where call recordings are stored.

Siri has also been redesigned, with a new indicator forming a multicolor border around your device to indicate a shift towards Apple Intelligence. In time, Siri will be able to understand natural language in case you stumble over your words and pull personal context from multiple apps like Mail, Messages, Contacts, and Calendar to answer specific requests. Additionally, Siri will be able to provide steps on how to complete actions sourced from device user guides, complete requested tasks based on information presented on your screen through ‘onscreen awareness,’ and complete actions in and across apps.

More broadly, Siri and Apple Intelligence can do all of this through what Apple calls personal context. Personal context is everything your Apple devices know about you because they have access to essentially all of your data. Apple Intelligence is unique compared to other AI tools, as those tools may know a lot about the broader world but little about you specifically.

Frankly, writing out all of the improvements being brought on by Apple Intelligence is kind of exhausting. There are many facets to Apple Intelligence. There are the features I just listed and the ways in which that data is processed. It’s how data is being processed that Apple hopes will be a significant competitive advantage. Apple aims to do as much processing as possible on-device, which would explain why Apple Intelligence is limited to M-series devices and the iPhone 15 Pro, as those all have 8GB of RAM, seemingly the minimum for Apple Intelligence to run effectively.

For situations where your request needs additional computing power, Apple offers private cloud compute. Private cloud compute is a custom-built server farm with specialized chips that securely processes these requests and deletes them once they are completed. User data is never retained.

Finally, for information outside your personal context, Apple intends to rely on third-party AI providers, starting with OpenAI’s ChatGPT, for what the company calls “world knowledge.” While Apple hopes to add more AI vendors in the future, the company’s deal with OpenAI to include ChatGPT natively into its operating systems comes with assurances that user data isn’t collected or retained beyond completing the request.

What’s missing from Apple Intelligence?

I hope calling features are next

While I say I’m a dedicated Apple user, that doesn’t mean I can’t appreciate what the competition is doing. For the most part, I think many of Apple Intelligence’s capabilities are competitive with what Google is doing for its Pixel devices. Understandably, each company has a different focus. Google has also spent much more time integrating AI into its devices than Apple. Still, features like natural language search for photos and videos and the new Clean Up tool feel competitive to Pixel’s screenshots search and Magic Eraser.

Nevertheless, where I want to see Apple catch up is with the Pixel’s calling features. It’s incredible how a Google Pixel can wait on hold for you or ask a caller who they are and the reason for their call. Equally, giving Siri access to world knowledge without relying on a third-party integration with ChatGPT would make the voice assistant more competitive with the likes of Google Gemini, which is renowned for its ability to answer general knowledge questions.

This is not to say that Apple should abandon its privacy-focused approach to AI. I believe one of the major reasons Google is able to develop these features is because of the massive amounts of data the company collects. The Pixel’s Direct My Call feature allows users to share their call data with the company to improve the feature for others and future use. At the same time, Wait Times is built on sharing crowdsourced wait time data.

My thoughts on Apple Intelligence

While I see the future, it also worries me

It’s clear to me from Apple’s marketing that the company doesn’t want us to fixate on the ChatGPT integration or private cloud compute but solely on the company’s focus on personal context. Overall, these new tools powered by Apple Intelligence aren’t bad. But they aren’t perfect either. I’ve heard a lot of commentary around Apple Intelligence, many muted being underwhelmed by the Writing Tools or summarization, some excited about transcription, and others downright opposed to the image generation features.

On the one hand, I can see Writing Tools being useful in many contexts, including possibly being a free Grammarly replacement. Moreover, as a person who frequently stumbles over his words, Siri’s ability to understand natural language and overcome flubs in speaking is highly anticipated. It’s just disappointing that Siri isn’t receiving those same improvements on the Apple Watch or HomePod.

On the other hand, while Apple restricts image generation to specific art styles, it can still be uncomfortable to know that anyone you communicate with can generate a custom cartoon version of you. I wouldn’t be surprised if this tool is used to bully others. Equally so, I’m not convinced that I want to trust Apple Intelligence to filter out and prioritize my notifications should something actually important get missed.

Then, of course, there’s the concern of how there’s no clear indication of whether your Apple Intelligence requests are being processed on-device or being sent to Apple’s private cloud compute infrastructure. Even with the company’s assurances that none of your data is being stored or logged, I’m sure many would appreciate being able to turn off sending data at all or at least some indication of when their request is being sent off the device. This is all without touching the idea of having ChatGPT or other AI providers directly integrated with your operating system, which I don’t particularly love myself.

When it comes to privacy, Apple can, and should, continue to press its competitive advantage in this space. It’s clear that Google, Microsoft, and OpenAI won’t take that approach unless a giant like Apple pushes them. At the end of the day, it’s not surprising that Apple entered the AI race. The tech giant didn’t have much choice as it faced competing solutions. Still, that doesn’t mean that Apple Intelligence can’t be improved.

As Apple Intelligence stands today, it’s a fascinating addition to iOS 18, iPadOS 18, and macOS Sequoia.

However, if Apple wants to keep its competitors at bay, it needs to continue to push Apple Intelligence with an emphasis on transparency while giving users more control over how their data is processed. Apple also can’t stop pressing down on the accelerator as it’s already behind and needs to push hard to rival the features competitors have and will release. Apple doesn’t need to be the best at AI. It just needs to make it useful.