(Credit: Apple)
Last week I was lamenting the current state of Siri, but Apple's WWDC keynote gives me hope that a revitalized version of the digital assistant will actually be useful thanks to Apple Intelligence.
When it lands this fall, Siri will feature a completely new look, swapping a pulsating ball for a glowing on-screen takeover. Calling it a "major step forward," Apple’s Kelsey Peterson, Director of Machine Learning and AI, said the new Siri is “more natural, more contextually relevant, and more personal to you.”
Apple Intelligence is basically Apple’s home-grown generative AI model that’s designed to run locally on your device for many functions and on a secure cloud compute for more advanced tasks. According to Craig Federighi, SVP of Apple Software Engineering, you’ll even be able to tap the best-known generative AI model, OpenAI’s ChatGPT-4o.
The key points of Apple Intelligence are that it’s personalized and private. So, it takes all the information on your phone and uses that for context when fufillling your requests. The newer large language model (LLM) backing Siri should also mean that it better understands what you’re asking. It works "even if I stumble over my words," Peterson noted.
Visually, the new wraparound glowing light somewhat resembles how Google Assistant looks on Android device, though that’s just around the bottom of the screen. The new Siri icon, which has barely changed since the voice assistant launched more than a decade ago, is a clear sign from Apple that this is a new era for Siri. It gives the company a way to remove the bad associations with a Siri that so often misunderstands requests.
Another interface change I welcome is that you can type requests into Siri now. For the longest time, Siri was the sole voice-only AI assistant. Microsoft and Google embraced this long ago on Cortana and the Google Assistant, respectively. Now, even the gesture to bring up the Siri keyboard resembles Android’s bottom swipe up to invoke Google Assistant.
Context Is King
One of the things that impressed me most when first using ChatGPT and Copilot was that you didn’t have to repeat information if you wanted to ask a follow-up question. For example, if you ask “Who was the king of France in 1623?” and then followed that with “What about in 1705?” the old-school “AIs” like Alexa and pre-Apple Intelligence Siri would be flummoxed. It seems like a basic capability, and the ineptitude of earlier-generation AIs is frustrating.
Apple’s goal for Siri is not just to understand the context of your previous interaction, but of everything on your phone. Of course, this will likely only apply to info in Apple apps like Contacts, Weather, and Mail, but the company is making APIs called App Intents for third-party apps to tap into Apple Intelligence. I can imagine asking Siri to play that song you heard two days ago on Spotify with the lyric “you make me complete.”
Siri will understand the context of what’s on your screen. This ties in with a goal of the Siri remake to give it new powers to take actions in and across apps. For example, when you’re in Messages and the person you’re texting puts a mailing address in the chat, you can tell Siri to add the address to that contact’s details. You could just as well tell the AI to add a picture in Photos to an email to dad, or send an article from the Apple News app to a text message.
Not Just for the iPhone
Siri runs across Apple’s other platforms as well, though the company focused on the iPhone in its WWDC keynote. If Siri and Apple Intelligence are embedded into macOS the way they are in iOS, that will draw a striking distinction between the way Microsoft added Copilot to Windows. Redmond mostly added the AI as a separate tool for generating text and images that’s not deeply embedded in the system, though it does analyze Edge browser content. This could change with the Copilot+ PCs coming out, but at this point Apple Intelligence seems more embedded into the OS than Copilot.
Apple Intelligence is also more restricted than ChatGPT, Google Gemini, and Copilot. This is vintage Apple: The company wants to make sure the new feature only provides what it deems useful and safe. For example, Apple’s AI can only create drawings, not photorealistic images such as those you can get with the other major gen-AI tools.
What Service Is Handling Your Data?
One of the things that will take a while to parse is which parts of Siri and Apple Intelligence’s new generative AI will be handled locally on the device, which will go to Apple’s secure cloud computing server farms, and which will land in ChatGPT’s Azure servers.
That last one should be the clearest of these, since Apple at WWDC stated that any time you call upon ChatGPT features, for things like advanced image or text generation, you’ll have to grant permission. But it’s less clear which functions will be handled on-device and which go to Apple’s Private Cloud Compute.
There's also a question of whether Apple will use OpenAI's ChatGPT technology within Apple Intelligence at some point, but for now Apple seems to be signaling that AI (Apple Intelligence) is all in-house technology, though in-house likely includes previous AI firm acquisitions by Apple.
What Do You Need to Get the Newer, Smarter Siri?
By the way, as with so many new Apple features in the past, you’ll need one of the latest and most expensive devices to get any of this functionality. Apple Intelligence will only work on iPhone 15 Pro and Pro Max (and presumbly, the iPhone 16), and you’ll need an M1-based iPad or Mac. If you’re lucky (and deep-pocketed) enough to possess one of those devices, you can look forward to the new smart features coming in beta form this fall. According to Peterson, “over the course of the next year, we’ll be rolling out more features that make Siri even more personal and capable.” Welcome to the brave new world of generative AI, Siri, the butt of jokes no more!