Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Apple reveals new AI architecture built around Google Gemini models

macrumors.com

Post reply on HN
Print view View original

361–370 of 609 posts

Re: Apple reveals new AI architecture built around Google Gemini models

#361

Post by Gigachad »

taneq wrote: post #218
sterlind wrote: post #26

Earlier quoted context omitted.

so to translate: - Apple has powerful capabilities in iOS to enable Siri AI. - EU's DMA requires them to allow users to install third-party AI backends. - Apple doesn't think parties other than themselves should be trusted with those iOS permissions. I guess it'd be like if Apple allowed a first-party screen reader for iOS, so they refused to allow third-party screen readers.

> Apple doesn't think parties other than themselves should be trusted with those iOS permissions. I think we have ample evidence that regardless of whether Apple in particular is to be trusted, tech companies by default are certainly not. Opening up access to users’ private data requires not just any given app to be trustworthy, but all of them.

In this case I think it is true that 3rd parties can not be trusted with this capability, but Apple brought this on themselves by creating a ton of other capabilities like AirDrop, Airpods integrations, and apple watch capabilities which would have been safe for 3rd parties to use but keeping them locked down so you'd get a better experience with Apple accessories.

Re: Apple reveals new AI architecture built around Google Gemini models

#362

Post by vunderba »

fastball wrote: post #356
jiggawatts wrote: post #352

Earlier quoted context omitted.

Only if you consider Google Image Search and Google Nano Banana to be "the same thing" since they both produce an image based on text input! Similarly, Google Translate's millions of lines of hand-rolled code has been entirely superseded by LLMs that do a vastly better job. The LLM-based AI assistants are based on a wildly different technology stack with very different capabilities compared to the legacy "if-then-els…

Was Google Translate millions of lines of hand-rolled code? The Transformer architecture was invented for Google Translate, before it was used to build "LLMs".

I don't know about millions of lines of code, but Google Translate existed WELL BEFORE transformer architectures and relied on more traditional statistical machine translation techniques. They later moved to a neural machine translation technique, and then only after that in ~2019/2020 swapped to transformers.

Honestly a lot of us who worked in the translation sector remember NMT as being a huge step up and in some language-pairs even surpassing DeepL at the time.

https://en.wikipedia.org/wiki/Google_Neural_Machine_Translat...

Re: Apple reveals new AI architecture built around Google Gemini models

#363

Post by zelon88 »

btown wrote: post #334
impulser_ wrote: post #267

Earlier quoted context omitted.

They are using Google Cloud. https://security.apple.com/blog/expanding-pcc/?linkId=100000... "Now, we are collaborating with Google and NVIDIA to run new Apple Intelligence workloads on Google Cloud, extending our industry-leading PCC privacy commitments to third-party data centers for the first time."

Per that link: I think there's an interesting question about whether a nefarious actor who's infiltrated a cloud provider with physical access to machines that are running signed operating systems, with signed binaries, with TDX remote attestation, and with hardware supply chain verification, has the ability to break the privacy guarantees of a tenant with Apple's sophistication. Certainly, one could tamper with the…

Spoiler alert; Google is the nefarious actor.

Re: Apple reveals new AI architecture built around Google Gemini models

#364

Post by tibbar »

airstrike wrote: post #315
materielle wrote: post #309

Earlier quoted context omitted.

That’s not so special, though? There’s a difference between Google infra running Google services. Versus any F500 company running their services on GCP. It’s a bit whacky to think about because Apple will operate Google owned software on GCP. But it should be sandboxed just the same. I’m not making a normative privacy argument here. Just pointing out that this is cloud business as usual. Perhaps it’s interesting Appl…

I think the difference is scale. This is Apple, so it's an enormous amount of devices. And it's a seamless experience, to the user, going from local model to cloud models. So the question about which model Apple was going to use and where has been highly anticipated, especially by the likes of OpenAI and Anthropic. Imagine if either one could say they have Apple as their customer? Apple certainly has the cash to burn…

[dead]

Re: Apple reveals new AI architecture built around Google Gemini models

#365

Post by deepfriedbits »

kube-system wrote: post #78
deepfriedbits wrote: post #44

Earlier quoted context omitted.

Bingo. They see this as the future commodity it will be. Customers will choose AI providers much way they choose a car: taste, price, a few other factors. And to your point, Google has a massive balance sheet, produces their own AI chips, and is not going anywhere anytime soon.

People are not even going to choose their AI providers in the future, it's going to be a part of some other product.

I could totally see that

Re: Apple reveals new AI architecture built around Google Gemini models

#366

Post by appplication »

SoftTalker wrote: post #351
btown wrote: post #334

Earlier quoted context omitted.

Per that link: I think there's an interesting question about whether a nefarious actor who's infiltrated a cloud provider with physical access to machines that are running signed operating systems, with signed binaries, with TDX remote attestation, and with hardware supply chain verification, has the ability to break the privacy guarantees of a tenant with Apple's sophistication. Certainly, one could tamper with the…

Why bother with all that cloak and dagger stuff when they can just buy the data? You believe Apple and/or Google isn't selling it? I have some land in Florida I'd like to talk about.

Having worked at Apple, I will say I firmly believe they do not sell data. I worked in data science and we had the shittiest inference because we had essentially no access, even internally, to longitudinal or cross-app user data. Best we had was 15 minute rotating sessions for a single app. There are internal teams dedicated to deanonymizing data to try to narrow down users - if they can successfully do so, and relevant fields that lead to deanonymization get permanently purged from internal logging.

I can’t speak to the current architecture but Apple has shown a consistent willingness to sacrifice access to user data in the name of selling privacy instead at a premium price (you could argue precisely because no one of their competition have any meaningful posture on this). I do believe they are quite serious in their commitment to that, as they have found this strategy to be more valuable than the data itself.

Re: Apple reveals new AI architecture built around Google Gemini models

#367

Post by conception »

Bud wrote: post #350
chvid wrote: post #338

Let us hope the EU forces Apple to allow the end-user to choose the external model: Wouldn’t it be amazing having privacy first local models calling out via a welldefined open protocol to a model of your choice: Claude, Grok, DeepSeek? Sounds like OS architecture done right - screw the kickback business model.

Let's actually not hope that, and let's not indulge the EU regulators' fantasies that they get to dictate the product design of products from non-EU countries. Force Apple to support all kinds of arbitrary models? That's a comically bad idea.

“Let’s not indulge US regulators fantasies that they get to dictate the product design of cars from non-US countries.”

Of course governments have the ability to decide what products are sold in their countries and how.

Re: Apple reveals new AI architecture built around Google Gemini models

#368

Post by chvid »

Bud wrote: post #350
chvid wrote: post #338

Let us hope the EU forces Apple to allow the end-user to choose the external model: Wouldn’t it be amazing having privacy first local models calling out via a welldefined open protocol to a model of your choice: Claude, Grok, DeepSeek? Sounds like OS architecture done right - screw the kickback business model.

Let's actually not hope that, and let's not indulge the EU regulators' fantasies that they get to dictate the product design of products from non-EU countries. Force Apple to support all kinds of arbitrary models? That's a comically bad idea.

Why would it be ok that a monopoly business prevents free competition and consumer choice by only allowing certain or a single model provider that likely gives them kickback via some opaque business deal?

Re: Apple reveals new AI architecture built around Google Gemini models

#369

Post by impulser_ »

zelon88 wrote: post #363
btown wrote: post #334

Earlier quoted context omitted.

Per that link: I think there's an interesting question about whether a nefarious actor who's infiltrated a cloud provider with physical access to machines that are running signed operating systems, with signed binaries, with TDX remote attestation, and with hardware supply chain verification, has the ability to break the privacy guarantees of a tenant with Apple's sophistication. Certainly, one could tamper with the…

Spoiler alert; Google is the nefarious actor.

I think the last thing Google wants to do is get on the bad side of their largest partners.

Re: Apple reveals new AI architecture built around Google Gemini models

#370

Post by ProAm »

janalsncm wrote: post #235
dofm wrote: post #223

Earlier quoted context omitted.

Is there a meaningful Google-Apple boundary in operation? They are buying the right to distill their own Gemini models and run them in their data centres (or at least data centres they control); unless I am missing something, this isn't going to be infrastructure that Google has operational control over.

If Apple is running the inference from Apple iPhones and Apple data centers then Apple has operational control. Google’s influence ends the moment they hand the weights over to Apple.

But google is paying SpaceX/xAi for compute... so...
Post reply on HN