Tech products get delayed all the time, but Apple's recent acknowledgment that the next generation of Siri is delayed -- with no timeline beyond "in the coming year" -- is uncharacteristic for the typically secretive company.
Even more unusual is that the problem isn't just with Future Siri. The current version of Apple's assistant seems to be regressing. According to a Reddit thread (and reported by Daring Fireball and MacRumors), asking Siri "What month is it?" results in "I don't understand." I got the same response using both Siri under iOS 18.3.2 and the new, glowing Apple Intelligence-powered Siri in the latest iOS 18.4 beta.
Siri with Apple Intelligence (left) and without (right) can't answer the simple question of which month it is.
Parsing a basic question like that doesn't seem to be a heavy request. Perhaps it never came up because it's the kind of question only someone waking from a coma or being rescued from a deserted island would ask. Still, making the same query on a Pixel 6 Pro brings up the current date (which, to be fair, is not exactly the answer I was looking for, but at least it presented the right information).
All of this is frustrating for shareholders, journalists (though we've grudgingly gotten accustomed to it) and customers, especially when they expect a level of assistant competency from Apple that just isn't there. And the secrecy invites the same kind of months-long drumbeat of "Apple is falling behind on AI" that led up to the reveal of Apple Intelligence. It could also explain reports of a(nother) Siri executive shakeup within Apple.
By taking the unusual (for Apple) step of responding to investor and media pressure -- and announcing features that aren't close to ready -- the company may have made things worse by confirming that analysts, reporters and fans were right.
The smart play would be for Apple to adhere to its secretive ways, not previewing its features and capabilities until they're much closer to being ready to ship.
Apple's inflated expectations
Apple's approach to product development has been to work on projects secretly, over years if needed, until they're ready to see the light. They're often not 100% baked at release, but when they're ready to be introduced to the public, the core features and functions are there.
I could cite plenty of examples. It's a valid argument that the Vision Pro is not a successful product -- it's expensive, it hasn't been broadly adopted by customers or developers, it's uncomfortable and so on -- but the essential elements such as processing power, micro-OLED screens and VisionOS are all there as a solid foundation.
When a product's existence is heavily leaked ahead of time, Apple typically unveils a finished version -- even if it's still limited in functionality. It was generally expected leading up to Macworld Expo in 2007 that Apple would announce a phone -- particularly following the embarrassment of the Motorola ROKR E1 phone. But no one expected it to break from other smartphones of the time with its large screen, lack of physical keyboard and full web browser.
When Steve Jobs revealed the original iPhone, it was a radical idea to not include a physical keyboard.
What's different this time is that Apple's promise for an advanced Siri, to anchor Apple Intelligence, seems to be in reaction to investors, the media and early-adopters obsessed with not just the presence of AI but also the immediacy of AI. Apple needs to be seen as an active player in the AI space with competitive features -- and that those are just around the corner.
Also around that corner? The yearly iPhone refresh. Apple, like other phone-makers, sees AI as an important driver of new phone sales, since only its iPhone 15 Pro and the 6-month-old iPhone 16 series models have the processing power to run Apple Intelligence. And that's how we got a WWDC keynote in 2024 focused on Apple Intelligence and promising that very soon Siri would become an intelligent agent that can pull data from every corner of your iPhone to respond to queries such as "What time does my mom's flight arrive?"
LLMs don't follow a traditional release model
Large language models (LLMs) such as ChatGPT are advancing at a record pace. They're now much more naturally conversational and can summarize large amounts of information well. Real-time audio transcription, for example, is game-changing for someone like me who has always struggled to hand-write notes.
At the same time, these AI technologies are not making the kinds of gains that tech giants like Google and OpenAI expect. Apple isn't the only company hanging its AI future on intelligent agents that know everything about us.
Perhaps Apple, like Google, saw the brain-bending pace of advancement in LLMs' capabilities and figured the bumps and stumbles it's facing now could be solved with a few quick bug fixes and AI model recompiles. With those smoothed over, connecting the pieces and presenting them as the next generation of Siri would take a few months.
But that's not how it's playing out. AI hallucinations and bad data are still a problem -- are you getting your recommended dietary requirement of rocks?
I suspect Apple is smarting not just from having to delay its Siri plans, but from being forced to do so publicly. And yet, even if Future Siri doesn't make an appearance in the near future, there are plenty of opportunities coming up to continue improving Apple Intelligence features. Work on iOS 19 and iPhone 17 models, plus preparations for WWDC 2025 are no doubt well under way. Now that there are fewer expectations for the stalwart assistant, perhaps Siri's year will improve from here.