So many resources dedicated to questionable products. Car. AVP. Meanwhile Apple’s done comparatively little in AI compared to a lot of other big tech companies.
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 and Meta Reportedly Discussed AI Partnership for iOS 18
- Thread starter MacRumors
- Start date
- Sort by reaction score
…because Apple doesn’t have a chatbot?
I really question whether people even remotely understand what AI is at this point.
I've personally been eyeing a light phone on and off, it's interesting watching how "dumb" phones have adapted.
Consultant groups steer many corporate strategies, whether you think the ideas are beneficial or harmful. Though their level of control is obscured, it's significant.
How was Apple ever going to do AR glasses which is supposedly a goal of theirs without AI?
Can you elaborate on what you mean here?
Apple has been using AI for the better part of a decade. What they're not doing is wrapping it up into a *chatbot* (which apparently is the new definition of AI?) to inflate a stock bubble.
I'm not sure what you're getting at with the link between glasses and AI?
Can you imagine if Steve Jobs used someone else’s data, like releasing a phone mapping app using 3rd party data.
Last edited:
ChatGPT processing is not done on private cloud computePlease rewatch the Keynote, then the SOTU, to learn how Private Cloud Compute actually works.
Blog - Private Cloud Compute: A new frontier for AI privacy in the cloud - Apple Security Research
Secure and private AI processing in the cloud poses a formidable new challenge. To support advanced features of Apple Intelligence with larger foundation models, we created Private Cloud Compute (PCC), a groundbreaking cloud intelligence system designed specifically for private AI processing...security.apple.com
Private Cloud Computer *calls* ChatGPT.
Order of operations:
1. Phone determines on device if its models can fulfill the request: No
2. Phone determines if PCC can fulfill the request: No, offer to search ChatGPT (curently). If yes, proceed
3. Query is made FROM PCC with all of its privacy protections in place TO ChatGPT. No personal information from the device originating the call to PCC is passed to ChatGPT.
Do you understand? Do you think this AI is just like SmarterChild from the AIM days?? Large companies are starting to use AI as part of their work streams. I know mine is. And AI is only going to get more advanced from here. Apple is seeming more and more like MS 20 years ago. Not pushing the envelope effectively. The investments that are made, not panning out. AVP could end up just like Zune.
Meanwhile, my employer has all the generative AI options domain-blocked and has copilot disabled in Windows and Edge via group policy.
Hospital.
Last edited:
I think you have little understanding of what it actually takes to improve AI from where it is today. Diminishing returns on the chatbot and image generation fronts are the only future here as the amount of data required to get to "the next step" of progress *simply doesn't exist*.
The internet has been mined, completely. Synthetic data won't work, because it's susceptible to hallucinations, so garbage in garbage out. Windows Recall was a veiled attempt at opening up a new vector of data to train on, but that's been canned. Google and the others are also trying desperately to find new data to train on.
If you want to learn about the real state of AI in the coming years, do some research on model decay, how much unreturned capital has been dumped into it thusfar, how much energy all of this takes to do, and tell me how any of this shakes out in a capitalistic economic model that doesn't have the patience to allow anything but short term self defeating approaches (subscriptions) to AI, are going to work.
There is no long turn return on any of this. Much of what exists today can and will be used in enterprise, some disastrously and some to good effect.
TLDR:
Key terms to look into and add up:
Data Scarcity
Model Decay/Drift
AI training energy costs
AI training capital expenditures
AI return on investment
None of this adds up to the vagaries that conmen like Sam Altman are promising, not in this economic model.
Last edited:
You’re clearly not a fan of AI if you think what Apple is doing is revolutionary
I think OpenAI is expecting to convert a good number of people into paying chatGPT customers. Being the only external AI company with this OS-level integration is a pretty powerful endorsement to the average person, and this deal is putting chatGPT in front of a huge audience that might not have used it yet.
If they can convert enough people, that’s what pays for compute. I don’t know enough about their compute costs and conversion rate to decided whether this is a stupid move or not yet. I’d imagine turning on the fire hose of iOS users all at once is going to hit them pretty hard and be quite expensive, at least for a few weeks.
What other company has a *personal* AI model?
As a neurodivergent, Apple’s simple demo of “what time does mom’s plane come in” was a breakthrough moment for me.
I expend most of my mental energy keeping my professional life in check, on my personal life side of things I simply cannot keep my commitments in order in a calendar. My brain just doesn’t work that way. So Apple being able to surface my mom’s flight from wherever I scribbled it down previously (email, notes, in a text, calendar, reminder, wherever) and present it to me, yea that’s a big friggin deal that literally no other company in the AI field can do right now.
To put it more simply, Apple packaged AI in a way that users don’t even have to think about. Meaning it’s going to be the most used AI platform by a long shot in just a few years because regular people don’t have to think about it as being an AI. It’s just a part of how their phone does stuff.
Most techies here seem to have an aversion to technology that makes normal people’s lives simpler, which I really don’t understand because that has *always* been Apple’s greatest strength.
You may *want* to type into some chatbot for work activities, but why would you want that kind of interface to use AI in your personal life?
Last edited:
I wouldn’t bet any money on that. ChatGPT is the only external AI vendor today, simply because the deals aren’t finalized with other players. Craig and Joz explicitly said they view these integrations like they do for choosing your default search engine in Safari. There will be more, and likely before ios18 actually ships…
Rather than Apple using AI to enhance the Mac experience in a substantial way. They may go the same way they did with search tools or browsers, let some other company put it as the default on a Mac in exchange for Apple getting paid a lot of money to not bother implementing a solution themselves. They'll make it secure only because they don't want a competing company running wild on their hardware. Personally, I think they should invest all they can in making AI enhanced programing tools for every one of their new OS's. That way, it can speed up the generation of an app marketplace. It would also help if these AI tools made porting software from other platforms much easier. If it spurs more development it will have been worth it.
I only recently updated some of my friend's contact pictures from way back then when the FB integration assigned them. And I haven't even had FB in like 10 years now 😄
The entire first hour and a half was Apple dramatically enhancing the Mac and iOS experiences. There was a clear delineation when the "not Apple" stuff came into focus.
I'm genuinely curious how you came to that sentiment that's bolded?
Sure, I know they’re looking to have other models available, but who knows when those deals with be finalised. Like I said though, I don’t know enough to know if this is a stupid move or not.
Preaching about privacy and then wanting to form a partnership with Meta - my kind of humor! 😂
I think it's genius. Meta and MS's shareholders are eventually going to be out for blood looking for a return on the tens of billions of dollars they've poured into their AI offerings. MS is already desperately trying to get businesses to add ANOTHER $30 per month for the privilege of using CoPilot (we laughed at our reps when they brought it up. Double our spending for something that can't even be used in our field? Why?).
The problem is that none of these companies have a business model to recoup the tens of billions they've spent the last few years pumping into AI training hardware (and all the associated costs of data centers for it). There is no viable ROI right now.
Apple has chosen to allow users to work with these models, but they're smart enough to look long term and realize that only the "Too big to fail" players are going to be around in a couple short years. Small AI startups, funded by SPACS (read, short term scams) will either fold or be acquired by the big guys. But the big guys still don't have a viable business plan to recoup their original investments. So by not partnering internally with AI companies, Apple can let the list of vendors expand and eventually collapse, while 90% of what outside vendors were called upon for will eventually be fulfilled by Apple's own internal offerings.
Apple is playing the long game on AI, and in the short term covering what most *regular* people would use AI for in the first place.