Siri to Become a Real Chatbot?

Article Comments
1

Originally published at: Siri to Become a Real Chatbot? - TidBITS

Mark Gurman of Bloomberg has reported on two phases of Apple’s Siri plans. First, to make good on its Apple Intelligence promises from 2024, Apple will replace Siri’s creaky command parser with a large language model based on Google’s Gemini. This initial update to Siri will appear in iOS 26.4, likely due at the end of March or the beginning of April. In theory, it will provide the “more personalized Siri” that can integrate your email, photos, and messages into its actions and responses. It will also include the App Intents system, which lets Siri control third-party apps based on your requests, and should be better at finding current information on the Web.

Gurman says Apple will unveil the second phase of Siri at WWDC in June 2026, when the company will build a voice- and text-based chatbot experience—codenamed Campos—into iOS 27, iPadOS 27, and macOS 27. Apart from Siri becoming a chatbot, those updates are reportedly more aimed at improving performance and fixing bugs, hopefully addressing the errors and excesses of Liquid Glass, Apple’s controversial translucent redesign.

However, even though the company is testing this new chatbot technology as a standalone Siri app, much like the ChatGPT, Claude, and Gemini apps currently available, Gurman says that the company doesn’t plan to ship that app. Instead, he says we’ll interact with Siri as we do today, which could undermine the entire concept of Siri as a chatbot.

I hope Apple plans some sort of persistent text interface, unlike the way Siri conversations disappear today. Such an interface becomes especially important once Siri can maintain ongoing conversations by responding to follow-up questions, comments, and commands. Chatbot transcripts are crucial for understanding what has been said and for formulating responses.

Additionally, although Gurman says Apple is considering limiting how much Siri remembers for privacy reasons, a conversation history is vital. I often search for and revisit old chats with ChatGPT—it’s incredibly helpful to refer back to previous details or to pick up where I left off without starting over.

Some have criticized the blank text box as an interface, but I’m unsure what they envision replacing it. Voice is an excellent add-on at times, but it’s basically still text and suffers from extreme linearity—you can’t easily edit or take back anything you’ve said using voice alone, whereas it’s trivial to edit or modify text in numerous ways before pressing Return. Even when I dictate, I often find myself editing the text to correct misrecognized words, particularly when they’re key to the query or conversation.

Plus, while Apple pioneered the move toward graphical interfaces, which excel at simplifying highly specific actions on objects or with controls you can see, power users often stick with text interfaces because they can operate on objects and metadata that can’t be seen. You can do a lot with files in the Finder, but you can’t rename files in a nested hierarchy according to a regular expression, something that’s simple with the command line. What chatbots bring to the text interface is the ability to be understood without hewing to a formal command structure.

One final concern: Gurman reports that Apple and Google are discussing hosting the new chatbot directly on Google’s servers rather than Apple’s Private Cloud Compute infrastructure. If that happens, it would represent a significant shift in Apple’s privacy stance—something worth watching as these plans develop.

5 Likes
2

One possible nuance is that this could be on Google’s servers the way that some of iCloud is on Amazon’s AWS, etc. If Apple is using end-to-end encryption so that Google cannot see the data stored on the servers, and uses something like iCloud Private Relay for all requests to these servers to protect client IP address info, this may not be so bad.

3 Likes
3

Well I already don’t use Siri, so no biggie.

2 Likes
4

Would agree with all of your concerns there Adam. I find the repository of chats with Perplexity to be key to its utility. Frequently I enter voice mode while driving and have a conversation about something I am mulling over, it’s good for fleshing out a topic. All there for me when I get to my desk and can whittle it to the useful points and follow up on citations.

5 Likes
5

I feel like Apple is lagging far behind on AI. Siri should have been one of the first AI chatbots out there. It is what Siri has always been missing.

The only time I use it is to make touch free calls when driving. But the opportunity is tremendous. Your whole life is sitting there on your phone. Calendar, contacts,messages and emails. Taken to the extreme, even your phone calls and conversations.

Imagine the usefulness it COULD offer.

Making a Dr. appointment and it could suggest the perfect time based on past history and not conflicting with your calendar or even your typical schedule.

Let’s you know when an oil change is near.

A quick note to let you know that Silver is extraordinarily high and since you mentioned that you inherited a bunch of coins from your grandfather in a text, you might want to sell them, and here are some places you could sell it.

The point being, Apple should have been leading on this, but they ALWAYS wait and follow afterwards. This would have been one where they should have lead.

At the very least by making Siri more chatty and useful.

On the other hand, Amazon is dropping the ball by making Alexa a premium charge so Apple may get away with being late. Amazon could have KILLED Apple Home if they didn’t get greedy on the AI features.

Disclosure: I don’t own any of these products. ;-)

6

I feel a similar way. Instead focus was diverted to a failed car project and overpriced VR goggles that failed to catch on, and more billions were poured into TV+/filmmaking. The result is Apple is now seriously behind with a lingering Siri left as lame as it is, and they’re playing catch-up to the point where they need to ask their main competitor for help while stuck with an underperforming streaming service (worse off even than Apple Music vs. Spotify) that’s neither making them big $ nor driving hardware sales. Missed opportunity. Tim might have been good about a lot of things no doubt, but under his leadership Apple completely missed what is perhaps one of the greatest tech disruptions (obviously not just LLM chatbots, but agentic AI and control in general) we’re likely to experience during our lifetime. I hope they now turn around and make huge leaps fast, otherwise the software side of the house will risk starting to resemble Nokia around 2008. Perhaps, the fake demo they staged at WWDC’24 will in hindsight turn out to be a good thing in that it forced Apple to admit they had blundered and then to finally commit to a serious game change. I’m hoping they’ll be successful.

2 Likes
7

I assume that we are going to have Siri in privacy mode or you just have to say “Siri, please forget the last 30 minutes.”

8

yeah me too :joy:

9

I’m not shy about saying my home is living in the Amazon/Alexa ecosphere. It’s not that I’m in love with the company and its business model. It was simply the first voice interface I could actually use with controls like lights and heating, and I committed to it before HomePod was a thing.

Alexa 2.0 or whatever you want to call it migrated to our devices over the weekend. It still has some rough spots—last night while it was responding to an impromptu question it said something like “Tone 1, the lights in the back office are on. Tone 2, would you like me to turn them off for you?”

But here’s the thing: when I asked it what “Tone 1” and “Tone 2” meant, it responded that those were inflection patterns it was supposed to use, and when I replied with the colloquial feedback “You said the quiet part out loud,” it responded with an apology and said it still had a few “rough spots” and would fix the issue.

It was also able to set up an “Alexa routine” to offer to turn on my studio lights when the Echo device in that room senses I’ve walked in, but not to do that unless I agreed. (I sometimes walk in for a moment to retrieve a device, so I don’t want the lights to come on unless I need them.) I did not have to open an app, or explain anything beyond my initial request/prompt.

This is what Apple has to contend with. They’ve had opportunities to use Alexa 2.0 as a premium product, and now it’s being released more generally. The things I see here:

  • In my household Alexa and Siri were both dumb robots responding to scripts triggered by voice commands. I would use Alexa for most of the home control, and Siri was on my wrist (Watch) if I needed a timer outdoors.
  • In my car Siri is much more useful for hands-free and eyes-on-the-road control, navigation, and communication. But it still just knows what it knows. Alexa’s not in the picture on the road.
  • Now, Alexa (2.0) suddenly has the potential to be crazy-useful. It feels flexible and responsive, and it has me pushing its boundaries. I reached the original Alexa’s boundaries pretty quickly, and its most-used response was “I’m not sure how to help you with that.” Alexa 2.0 has already demonstrated its capacity to generate small programs, interact in an intelligent way, and to get better with time.
  • That leaves Siri in the dust at home. I am more inclined to use the one Echo Dot speaker I have in a sheltered outdoor space for timing, versus Siri.

For people who don’t want to use automated assistants, this will mean nothing and move nothing. But for people who do, Alexa currently has no meaningful competition from Apple, and will not until Apple ships something that looks like an intelligent Siri.

Since they’ve been hanging so much prestige-capital on the Siri ecosystem for two years, I don’t blame them for looking for a shortcut like the Gemini engine.

10

A.I. is still in its infancy and is still advancing and improving. However, the hype doesn’t justify the progress. A.I. LLMs are faking it till they make it. Literally! I’ve spent many hours conversing with Chatbots, not a long running conversation either. The longer you go the likelier A.I. is going to start feeding you bad data and sending you down a fruitless rabbit hole. ChatGPT had some poor man thinking he solved some major mathematics problem. Once he reached out to a Phd, he was informed that it was all fantasy. That none of it works and that he solved nothing. Several people with mental illness had very bad results using ChatGPT. Their needs to be more constraints put in place on A.I. Chatbots. Also it would be ideal to make a law saying any images or animations created by A.I. have a visible watermark while posting on social media. So many morons believe it’s real.

A.I. is best utilized by experts in their field. You need to be an expert so you are capable of judging the A.I.’s response is incorrect. All these people Vibe coding without knowledge are going to get severely burned sooner or later. That’s not to say vibe coding is bad. But it’s not the final solution. You still have to go in there and fix the code by hand. LLMs are not software engineers. Not yet anyway. Microsoft claimed 30% of their code is A.I. well it shows. Have you seen the security update that broke calculator, notepad, and you couldn’t even shut down Windows? The Microsoft CEO made statements indicating he’s pulling the reigns on A.I. and he wonders when the bubble is going to burst. I’ve been in IT work since the 80’s. I’ve seen my share of ridiculous hype trains. This is no different. It’s a race to see who gets to create an Artificial General Intelligence first. Who will be the dominate A.I. that sets the standards for all the others. Trillions of dollars are in play. If they don’t start producing real world results that produce real productivity improvements and financial savings. Then it’s all going to fizzle out.

There will be a lot of money to be made fixing companies who broke themselves with A.I. by trusting the hype train. It is already happening. One company fired their marketing staff in favor of using an A.I. to rebrand their business. It failed miserably and then BBDO, the global advertising giant to fix it but at emergency pricing. This is clearly happening all over. If the hype doesn’t stop it will get insane really fast.

2 Likes