Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Adds Disclosure About Delayed Siri Features to iPhone 16 Pages
- Thread starter MacRumors
- Start date
- Sort by reaction score
Failures like Apple Lisa, Apple iPod Sox, and Apple HiFi?
Regarding Apple Intelligence... It's hardly a failure as it hasn't been brought to market yet.
You may not realize many product introductions occasionally suffer from introduction delays. All tech companies experience that.
That's certainly understandable considering Apple's *privacy focused* entry into AI. I would much rather have a delay on an Apple AI product than have it introduced prematurely with major issues - especially issues that are privacy related.
What's really funny is the same people who are unhappy about Apple's AI delay are the same people who vociferously claim in multiple threads they don't want Apple AI on their Apple products. And will quickly turn it off.
Why be unhappy with something delivered late that you don't want to begin with? And will immediately turn off? That makes no sense.
A massive class-action lawsuit is coming. Everyone who bought an iPhone 16 will be eligible for damages. They literally sold a non-existent product with fake features. Steve is spinning (again.)
I haven’t really followed Pei’s career, beyond picking up a pair of his original nothing 1 earbuds, but he does come across well in quite a few interviews I’ve read/seen.
And this comment is very much along those lines. You’d hope he will make the shortlist when Cook is eventually forced into retirement.
I recall the ads for the phone saying "Coming in a future software update" with no guarantee of the time of that so still is legally sound.
I suppose the truth is that personal context is easy to do if you give up privacy. However, Apple's stance on privacy means that they need to do it on device or via the private cloud, which is a lot harder to do.
Which goes back to my earlier point of why announce until it's ready. Apple has never been first to market. Heck - they didn't even announce iPhone until they were basically ready to ship. Why are they doing this crap early now? What's the rush to AI?
Which goes back to my earlier point of why announce until it's ready. Apple has never been first to market. Heck - they didn't even announce iPhone until they were basically ready to ship. Why are they doing this crap early now? What's the rush to AI?
Somehow I don't think privacy is the entire problem. For one, the personal context AI needs to scan all incoming messages/emails/communications for interesting tidbits and classify them appropriately.
So, if your dad is flying in next week, the AI knows it's a calendar event. And if your kid asks you to buy avocados on your way back from work, it's classified as a shopping list item.
You'd think a larger model running in the cloud might do a better job at classifying this stuff.
But as far as I understand LLMs construct their answers by finding out how strongly different bits of information are associated in the material they are trained on.
Imagine you ask Gemini the year Steve died. Gemini has been trained on a massive amount of data where the association of Jobs and 2011 recurs many times, and it's very likely to give you the correct answer.
Now imagine you ask the LLM what you need to get at the grocery store on your way back home.
Your kids asked for avocados.
Your wife asked for toothpaste, but she also asked if you want to get pizza on saturday night (today it's wednesday).
There's a party at the office and someone asked who's getting the cake.
Here you don't have a strong association that drives the LLM's answer. It has to pick avocados and toothpaste, ignore pizza. And maybe remind you that there's a question about cake.
One problem is how to let a LLM in the cloud securely scan all your info. But I think the other problem is that personal context kind of answers are harder for LLMs to build correctly.
Now, if someone with real experience on LLMs chimed in...
My view is that I spent $1200 on a phone that is supposed to have intelligence, now it’s not going to arrive until the iPhone 17 comes out.
It’s amazing that they launched ad campaigns, etc., around this when they had no clue when it would be done.
Fraud.
Disclaimer: I'm not an LLM/AI expert. In fact, I'd consider myself a bit of an LLM/AI novice. But I previously worked on AI (back when we called it ML) and have picked up some knowledge...
Basically, you're correct. But it's possible to give a higher weight to user submitted data, and rank it higher than generic training data. And if your wife asks for pizza on Saturdays, and the LLM can read that from your text message database, it can infer that on Saturdays, maybe you will want to get a pizza.
There's where the privacy nightmare starts:
Because not only can I ask it about dinner on Saturday nights. A malicious actor (or app with text message privileges) can ask it what I like to eat on which nights, and the LLM will happily tell them that my wife likes pizza on Saturdays with an 89% probability. e.g. Pizza Hut can then use that data to send pizza ads to my wife on Friday evenings or Saturday afternoon, ensuring that Saturday night we order from Pizza Hut.
The worst part is that you have algorithms mining my phone for my wife's data, and targeting her, based off of data I own. So every friend you talk to, is a potential vector for a privacy nightmare. It's not just your data you need to worry about, but your friends' data security too.
With everything having AI "cores" now a days, it's not computationally expensive to run the algorithm overnight. If the app that is profiling you doesn't need 100% accuracy, but is OK selling you ads based off of 85% accuracy, or even 75% accuracy, then they can use a much smaller model, meaning you might not even notice that your phone's apps are spying on you.
I spent the same $1200 and mine has useful, basic stuff, which serve me well.
There are laws in place to protect people from false advertising.
It’s not about things going wrong. They literally sold a device based on false information and lied to consumers.
Now they are removing all evidence from their site lol.
And absolutely nothing will happen as Apple stated its AI will be rolled out slowly once the remote server infrastructure is in place. In the meantime some things are handled on phone. And that works great for me.
And... All the people here are stating they don't want AI on their phone, and will be turning it off when it comes. So nothing to get upset about if it comes late. Nobody here wants it.
Basic stuff that the iPhone 15, 14 and 13 are all capable of. No need to buy the new “built for intelligence” one, obviously.
The summarized emails and messages, genmoji, and image playgroundare not basic stuff that can be done except on the iphone 15PM and newer.
All that stuff is a mess in my opinion, especially the summaries. Seems like every reviewer has turned the summaries off because they are so dumb and misleading.
Cheers.
It’s not impossible to promise something to someone whom doesn’t really know the process of AI and all it encompasses.
Tim Cook even says he delegates a lot of responsibility to his people, but he only deals with the vague details. So he has to trust someone almost blindly in this case as he is no AI expert and would understand what technical processes would be involved.
That said, it was damn concerning to show advertising on features not yet fully baked. This is AirPower all over again.
Tim Cook even says he delegates a lot of responsibility to his people, but he only deals with the vague details. So he has to trust someone almost blindly in this case as he is no AI expert and would understand what technical processes would be involved.
That said, it was damn concerning to show advertising on features not yet fully baked. This is AirPower all over again.
In this year's WWDC, Craig will introduce Apple Intelligence 2, which requires 16GB RAM and only available for device with at least 512GB storage space.