The vast majority don’t?
Expecting every device to work with the bleeding edge of AI is kind of a silly thing. Can’t exactly go back and change a chip from 4 years ago to support something tomorrow. Nor do you want to inhibit what you can do today.
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 Intelligence' Personal AI Unveiled for iPhone, iPad, and Mac
- Thread starter MacRumors
- Start date
- Sort by reaction score
Expecting every device to work with the bleeding edge of AI is kind of a silly thing. Can’t exactly go back and change a chip from 4 years ago to support something tomorrow. Nor do you want to inhibit what you can do today.
the topic is RAM. Apple was still shipping only 6gb in their flagship in 2022. Samsung was shipping 8-12GB back in 2020 in theirs.
working out great for Apple and their shareholders, though!
Terrible idea and opening massive privacy risk integrating chatgpt into siri, millions of people queries with potential sensitive info will be sent to openAI servers, and I'm pretty sure openAI stores all chat logs for further training.
I used chatgpt maybe couple times and found it useless (maybe only useful for cheating on schoolwork), this is adding a ton of useless bloat no one needs or asked for.
I used chatgpt maybe couple times and found it useless (maybe only useful for cheating on schoolwork), this is adding a ton of useless bloat no one needs or asked for.
Last edited:
The 15 Pro series is the only iPhone series, to date, with at least 8GB RAM.
Nobody bought the iPhone 15 Pro with Apple Intelligence either.
I actually like what I saw. Unfortunately, I’m not getting a new iPhone to try this out. Hopefully, people like it once they actually use it.
Wonder if HomePod will benefit at all from the improved Siri..? I can choose to not use Siri on my Macs, iPad and iPhone to keep my blood pressure down, but it's not as easy to opt out when using the HomePod.
Are you saying they manufacturer their own switches, as opposed to contracting others to build them??
Apples to Oranges. Different ram. Different architecture. Different language. Everything running on iOS is compiled languages like ObjectiveC and Swift. The memory footprint is exceedingly efficient and fast. Contrast to how much of the Android world runs in the JVM. Like saying an F350 is faster than a Mustang because it has more power.
And to your point, consumers don’t care about this. It has no bearing on anything. It’s easier to say AI works with iPhone 15 vs you need this CPU and this RAM.
They use an ODM. They use the same supply chain that makes their laptops to make servers and switches. The OS, design and firmware are in house. Even network cards.
None of the companies like Dell Cisco HP make their stuff. It’s all Quanta, Wistron, Huawei, Foxconn, etc.
Looks very good. Cannot wait to try it out. Glad that my 15 Pro Max is supported. Also not surprised that my 2020 iPad Pro is missing out.
It’s not that. You are using Apple intelligence. ChatGPT is for things that you would google. The Internal AI deals with privacy related queries. If you ask for like a recipe or how to make your concrete set faster, you can choose to send that to Chat GPT with a confirmation dialog.
Has no bearing on anything? You have one iPhone model that supports AI because... reasons.
Yeah, it's much easier not to explain how Apple being cheapskate screwed over their customers. Much less discomforting that way. Let's pretend it's just an unfortunate turn of events and not decisions made by Apple. And we'll never get any clarification or explanation from them, either, leaving people to argue in circles instead
I don't think that any company has transcribed user data into a vector database on the local device, then set up a simple LLM to selectively feed just the right vectors to a much larger LLM running on a cloud. That is not easy. That is not even close to easy. I have played with that myself. Doing so on a cellphone without having it melt or burst into flames is a significant challenge. Just converting the data into vectors could quite likely drain the battery in less time than it takes to sneeze.
I suspect that context memory will all be in the cloud. The local device will take local data, probably convert it into vectors, then feed the cloud hosted GPT instance the data that it needs. The remote host will run a Langchain tool that requests the data from the small LLM on the local device. That data will go into context in one way or another.
You need context memory to be on the device doing the actual computation.
Because then they would have to support the 14 Pro as well and that doesn't fit into Apple's scheme to force us to buy new iPhones.
There are open source topo maps and licensed ones. Anyway I think that Apple Maps already encompass many layers of proprietary data (for routes, traffic data, 3d buildings etc) so...if even third party apps in the Apple Store provide topo maps without breaking a sweat I really don't understand why Apple can't afford to reach a deal to embed topo maps for the whole Earth.
And it is Apple themselves who decided to not give the 15 the 8GB so I don’t get the point of this. Galaxy S10 had 8GB of RAM, 5 years ago, at a lower price point, and it was a beautiful device for its time that had no compromises when it comes to internals, camera, screen and so on. Apple makes the decisions for both the product and the software it runs, so why skim on the RAM when it's going to be an issue?
I'm not giving up my iPhone 12 mini for this Tim
Technologically Apple Intelligence seems to be mostly the same thing all other companies have been doing with large language models.
The biggest takeaways for me are:
1. There is a small model that will run on device for specific simpler tasks, which is good in my opinion.
2. Apple's "innovation" in this space is mostly about integrating it strategically into the OS.
One of the biggest problems of large models like ChatGPT 4 is, that it is very hard for a novice to discover what it is capable off. It's mostly just a text box and as a user you are supposed to guess what to do with it.
So my 14 pro max is obsolete but my M1 MBP which is nearly 2 years older is fine...