This is under the assumption that we don't see any significant optimization in the meantime. Looking back over the last eight years, the probability of no progress on the software side is near zero.
For a breakthrough in the consumer market, running LLM on-device with today's capabilities requires solving one key topic: "JIT learning" [2]. We can see some progress here [3, 4]. Perhaps the transformer architecture is not the best for this requirement, but it is hard to argue that it is impossible for Generative AI.
Due to today's technical limitations, we don't have real personal assistants. This could be the Mac for Apple in the AI era.
[1] https://gadgetversus.com/graphics-card/apple-a11-bionic-gpu-...
[2] Increasing context size is not a valid option for my scenario as it also increases the computation demand linear.
[2] https://arxiv.org/abs/2311.06668
[3] https://arxiv.org/abs/2305.18466
[Edit: decimal separator mess]
Audio, video, screen recordings, etc. from a single customer could be something between 1 and 10 GByte per day on average. After training you might get something like 3 MByte in additional model size per day. Even with 1 billion active users you would need to store additional data with 1 billion GByte (again on hot storage, like expensive GPU memory). The total amount of the memory of GPUs sold by NVIDIA is not even close to 400mio GByte (NVIDIA 3.8mio data center GPUs in 2023).
When distilled, most people’s days consist of very few actual newly discovered facts, decisions and changes to context.
To be clear, just having a chatbot website/app does not count.
- If you are wearing an Apple Watch a ML model is constantly analyzing your heart rhythm and will alert you to (some types of) irregularities. It's so computationally efficient it can literally do this in the background all day long.
- When you take any picture on any iPhone a whole array of ML models immediately run to improve the image. More models are used when manually editing images.
- After you save the photo ML models run to analyze and index the photo so it's easily searchable later. That's why you can search for "golden retriever" and get actual results.
- When you speak at your device (for example, to dictate a text message) there's a ML model that transcribes that into text. Likewise, when you're hands-free and want to hear an incoming text message, an ML model converts it to audio. All on-device and available offline at that.
Or are we playing that stupid game where "AI === LLM"?
Well, the original question was specifically about LLMs. ("What other companies have successfully integrated LLM tech in their mainstream products?")
Nvidia
Also, it sounds like Cook and Federighi just repeated talking points the public has already heard, so I'm not sure what the point of this was.
If there are any current Apple employees here, maybe they can weigh in.
Unless things have changed in the last 15 years, my understanding was that they actually are barred from doing just that
Naturally then you need the proof any random nickname is actually who they say they are.
I don't need proof. I know that Apple employees visit HN and sometimes comment. In any case, this is not a high-stakes product leak: I'm just asking for their personal impressions of this company-wide meeting that has already been reported.
Though it sounds like what they actually got was fairly in-substantive statements without a clearly articulated AI strategy.
Doesn't mean that Apple doesn't have a promising AI strategy though, if so, it wasn't communicated in this Pep Talk: so what was the point?
Perhaps to look like they are doing something? Are empty words better than no words at all?
That's the biggest shift I've heard from Apple. They were either "first" or ignored the existence of competing features/products for ages. I'm really surprised by this quote.
Compare "smartphones before iPhone" to the original announcement:
> iPhone also ushers in an era of software power and sophistication never before seen in a mobile device, which completely redefines what users can do on their mobile phones. (...) iPhone is a revolutionary and magical product that is literally five years ahead of any other mobile phone,
When the iPhone launched, the Android project changed direction toward a full screen phone and that form became much more dominant and popular than the BlackBerry form.
Apple made the bet that they could make the full screen experience much more compelling that people would accept the trade off of losing the keyboard.
Strongly disagree with this. Their marketing often claims inventing things that have existed.
Strongly disagree with this. Their marketing often claims reinventing things that have existed, or revolutionising them, or reimagining them, but rarely claims to be the first, ever, without qualification.
There were smartphones before iPhone, now all smartphones are black featureless rectangles. There were printers before LaserWriter, then for 20 years all printers became this. (And later disappeared.) There were wireless heaphones before Airpods, now the difference is in the shape of the stubs. There were laptops before the Macbook Air... etc
I recall marketing comparing iPhones to blackberries. They even had iTunes running on Motorola phone
https://www.makeuseof.com/itunes-phone-before-the-iphone-exp...
Nobody claimed Apple was the first at this. They were just the best, eventually. But it’s been 20 years
But they sure write releases like that's implied.
> iPhone introduces an entirely new user interface based on a large multi-touch display and pioneering new software, letting users control iPhone with just their fingers.
Large display existed before, "just fingers" control was... always the case, the interface was quite polished, but it existed elsewhere, etc. But if you didn't know that, reading the announcement sure sounds like it's never been done before. It's the multi touch that makes the combination novel.
There were smartphones before the iPhone. Consider the IPAQ and Windows Mobile 6.0.
And of course plenty of MP3 players before iPod.
While cool is subjective, what new, mass-market products should they create? Which product market should they "re-invent"? I wish they'd buy Sonos and fix that shit show, but that's not a profitable market to enter.
Barring the price jump around the iPhone 7, their smartphones have stayed about the same price. [1]
Over half of all smartphone repairs are battery replacements, which implies people don't take care of their batteries or are keeping their phones long enough to wear the battery down normally. [2] Additionally, Apple ranks very well in repairability. [3] They also support their phone's software longer on older devices than the competition.
[1] https://www.androidauthority.com/iphone-price-history-322149... [2] https://www.businessresearchinsights.com/market-reports/smar... [3] https://www.ifixit.com/repairability/smartphone-repairabilit...
It will certainly attract less talent, so even that is slow enshittification. The sane thing would have been to use the software ecosystem to sell the platform, and have no fees.
While OpenAI sells $2 bills for $1, Tim Cook was out there increasing service revenue and profitability so that it was larger than Macs and iPads combined.
Tim Cook presided over some incredibly lucrative product launches like AirPods, TV+, Apple Music, moved chip design in house which doubled Mac market share and has made the iPhone continually dominant, they’ll even drop third party 5G models soon. These are all incredibly shrewd long term strategy moves.
Excessive greed obliterates goodwill.
You don't seem to understand sj placed cool first, while TC was the bean counter and continues to optimize this while jumping the shark.
Current Apple has a deep, systemic lack of cool and lack of entrepreneurial leadership with good taste that will ultimately lose the crown.
M-series chips are insanely cool (literally and figuratively) and have no competition even five years in. Same with the W-series SoCs on watches.
Are there any third-party haptic vibration motors yet?
Shoot, even their trackpads, which already stood alone, have gotten _better_ over the years.
Nobody else will have their own vertically-integrated modem out in production. This will make budget iPhones (maybe all iPhones) so much cheaper once they show Qualcomm the door.
That's before the advances they've made in software, like their camera processing pipeline (which only gets better; their video stack still has no equal) and differential privacy.
Oh, yeah, and the Vision Pro, which basically everyone who has tried it has said that it is the most advanced technology they've ever used.
Almost all of this happened under Cook's tenure.
Apple is still in the business of building insanely cool shit.
Cool is the kind of thing people envy. It's the kind of product that gets namedropped in a music video.
iPhone will get some new exciting feature and everyone will wonder how they managed to do it at the price point nobody else can.
Last weekend my grade school friend visited and we took a boat out on Puget Sound. We followed two orca whales for 45 minutes and I shot multiple videos of them in 4k 60fps. They look beautiful and played seamlessly on my TV without any crappy ads. I shared them with our two families easily through airdrop and later through iCloud when we got back to land.
That evening my son was dancing and signing he favorite songs with my friend’s son so I shot a “music video” of them and shared it again.
Both these experiences were exciting to me. It felt insane to do such things with a pocket computer Apple hasn’t screwed up in nearly 20 years.
Like I said, it’s the whole product experience that’s the point. Apple has earned enough of my trust to believe one will come and even if not, I’m satisfied with what I have today and will have tomorrow.
Apple thinks/knows/gambles increasing vertical integration by building its own modems will make things even better.
I'm kinda jazzed about this.
Integration, power, and UX improvements, sure.
But (as a noob), I imagine having their own base band (?) chip will be like having built-in software defined radios. Opening up new applications and uses cases.
Cool is a vibe, not tech specs and little things. It's a whole aura.
Apple is not cool.
But those millions of data points can (rarely, briefly,) coalesce around a product or company, even though that’s mostly out of the control of those building the product or company.
EG if you asked someone in 1965 if a Jaguar E-Type was cool, or someone in 2000s London whether the Fruityloops DAW was cool, they’d say “yeah”.
I’m mostly agreeing, and it’s a super minor point, but tech specs are part of the unknowable, constantly-shifting constellation of symbols that produce “cool”, and there isn’t a reason an Apple product couldn’t, in the future, align the stars. They did before! The white iPod earbud wire did, briefly, signify cool.
Says who, exactly? It is very cool for most.
I generally agree with what you're saying, and unlike Cook I don't find it "hard to imagine" life without iPhones (an augmented reality future isn't that far out to consider), but they have a long runway. Gen Z and Gen Alpha define "cool", and they are committed to the ecosystem.
Steve placed “Steve likes it” first. Many of the things he liked were terrible.
Cool things happen(ed) at Apple in spite of executive leadership, not because of it. The better E-team players know to stand back and not get in the way.
If they push forward with local AI that can be somewhat trusted, it would be a huge win.
How is that touch screen?
They layperson thinks of Apple quite positively. Nobody cares what us nerds here think is cool. The bazillion people who happily bought AirPods don’t care that a tiny group of nerds are still mad that Apple took away their dork expansion slot and dweeb ports.
Apple is more trusted than any other tech company with privacy in consumer surveys.
Apple is the smartphone company with the highest customer satisfaction ratings (over 80%).
Consumer reports ranks Apple first in computer support customer service.
These are quantitative measures.
Apple very clearly has command of their messaging and there really aren’t any cracks in their strategy.
Recall Solyndra's fate.
A single US-based manufacturer with very modest govt support vs China's all-in strategic industrial policy, commitment to dominating the market, spinning up the entire supply chain composed of 100s of companies.
Its the same situation with batteries and automobiles.
In a different timeline, we'd have a "master plan" of how to ensure industry (and the populous) is able to thrive in the future. I believe it's technically a solvable problem but it requires adults in the room, who value the the country as a whole. We do not have that, and I'm afraid it's possible we never will.
It’s the kind of mistake an LLM would make. Very Lacanian.
Like Nadella, you need someone from the early years, who knows the business, to run it.
Cook lacks product vision because 1) he’s no Steve Jobs. He was hired. He didn’t create. 2) He doesn’t have an Jonny Ive to make something as boring as a computer be as sexy as an Italian vase. Or as sleek as a pencil. Or as flat as paper. Or whatever metaphors were used during the Ive years to describe his design process.
But he has been there long enough to know how it works.
Windows seems to be a platform for selling cloud services… lamentably it probably works.
https://next.content.town/p/unfucking-windows-for-fun-and-pr...
There’s a lot of settings and things you can regedit to make it great again, sane again, plain again.
Still not a customer facing / product development role. At the same time though, again, so much of what makes Apple's products so good is that they have been amazing at product having to work with manufacturing to push the bounds of what is possible. Apple Vision for example taps this intersection: part of the product very much was figuring out physically what it was you could build.
(Something about the past year has really really shifted my perspective, enhanced the already huge respect I have for people making physical things.)
Based on what exactly? He led the overhaul of a massive amount of Apple under his tenure.
I was leaning towards buying a Mac, but now I won't because I do what (little) I can to slow down AI.
Switching to Windows would also clearly be encouraging the AI juggernaut, so I will stay with Linux.
It's not my fault that the reality in which humanity find itself turned out to be more dangerous than almost anyone suspected. My only moral obligation is to do what I can to make the future turn out okay even though what I can do is very very little.
I think you're focusing on the wrong things. AI can be used in harmful ways, but not because they're outsmarting human beings despite all the cult-like hype. In fact, they don't need to be actually competent for the rich to take advantage of the tech in destructive ways. They just need to convince the public that they're competent enough so that they have an excuse to cut jobs. Even if AI does a poorer job, it won't matter if consumers don't have alternatives, which is unfortunately the case in many situations. We face a much bigger threat of data breaches from vibe coded apps than conscious robots manipulating humans through the Matrix.
Just look at Google support. It's a bunch of mindless robots that can kick you out of their platform on a whim. Their "dispute process" is another robot that passive-aggressively ragebaits you. [1][2] They're incompetent, yet it helps one of the richest companies in the world save money.
Also, let's not forget Google's AI flagged multiple desperate parents sharing medical pics of their kids to their doctors. Only when the media contacted them did a human being come out, only to falsely accuse the parents of being pedos. [3] People were harmed, and it's not because of competency.
Another greater concern is the ability of LLMs to mass-produce spam or troll content with minimal effort. It's a major threat to democracies all around the globe, and it turns out we don't need a superintelligence for demagogues to misuse it and cause harm.
There are more real concerns regarding AI other than the perpetually "just around the corner" superintelligence. What we need is a push for stronger regulatory protection for workers, consumers, and constituents. Not boycotting Macbooks because of AI.
[1]: https://news.ycombinator.com/item?id=26061935
The ability to have true personal AI agent that you would own would be quite empowering. Out of all the industry players I'd put Apple as the least bad option to have that happen with.
To be the least bad option, Apple would need to publish a plan for keeping an AI under control so that it stays under control even if it undergoes a sharp increase in cognitive capability (e.g., during training) or alternatively a plan to prevent an AI's capability from ever rising to a level that requires the aforementioned control.
I haven't seen anything out of Apple suggesting that Apple's leaders understand that a plan of the first kind or the second kind is necessary.
Most people who have written about the topic in detail put Anthropic as the least bad option because out of all the groups with competitive offerings, their leadership has written in the most detail about the need for a plan and about their particular (completely inadequate IMHO) plan.
I myself put Google as the least bad option -- the slightly less awful option, to be precise -- with large uncertainty because Google wasn't pushing capabilities hard till OpenAI and Anthropic put it in a situation in which it either had to start pushing hard or risk falling so far behind it wouldn't be able to catch up. Consequently, I use Gemini as my LLM service. In particular, Google risked finding itself in a situation in which it cannot create a competitive offering because it doesn't have access to enough data collected from users of LLMs and generative AIs and cannot get enough data because it cannot attract users. While it was the leading lab, Google was proceeding slowly and at least one industry insider claims credibly that the slowness was deliberately chosen to reduce the probability of an AI catastrophe.
I must stress that no one has an adequate plan for avoiding an AI catastrophe while continuing to push capabilities, and IMHO no one is likely to devise one in time, so would be great if no one did any more frontier AI research at all till humanity itself becomes more cognitively capable.
Maybe this will help: AI labs have tried out 100s of different designs for AIs and if they aren't stopped (e.g., by the governments of the developed world) they are going to try out 1000s of additional designs. Most of us who worry about AI takeover or about human extinction caused by AI do not claim to be able to tell which design will be the first design capable of taking over or of extincting humanity -- even if we had complete access to the source code and the training data and we could ask the researchers behind the design questions. (The researchers do not know either IMHO.) But once an design has been widely deployed for many months, we know that that design is not the one that is going to take over. Gemini 2.5 for example has been widely deployed since January. It has been given plentiful access to very gullible people, very desperate people and plentiful compute resources. (When a customer asks an AI to write code, then runs that code without first understanding the code himself, that is giving the AI access to whatever compute resource the code gets run on.) If Gemini 2.5 were able to take over the world, it would have done so already. Ergo, I consider it morally permissible for me to use Gemini 2.5. Now Google is not going to stop with Gemini 2.5: it will continue to try out different designs, which is why I consider it my obligation to avoid helping Google, e.g., by giving it money, which is why so far I've stayed on the free tier of Gemini.
Gemini 2.5 is not very agenty: it does not learn continuously, it is extremely unlikely that its can work effectively towards any long-range plan or devise a plan that can withstand determined human opposition. So in the particular case of Gemini 2.5 and its competitors, we really didn't need many months of wide deployment to go by before we can conclude with high certainty that Gemini 2.5 is incapable of taking over the world. But most AI researchers and most leaders in AI labs consider the fact that the current crop of deployed AIs cannot learn continuously very well and cannot formulate and work towards long-range plans as deficiencies to be overcome.
I should be able to replace Siri with any AI provider over a year ago. (Eg hold power button and immediately talk to gpt4)
How’s something so easy so hard for Apple?
The friction of one button press to 3-4 interactions is significant enough to make it not worth the action button, especially since holding the power button is supposed to do the same thing.
To detriment of Windows, XBox hardware, .NET team shooting into all directions.
Sometimes company’s just don’t do good enough.
It remains to be seen whether this was a smart move, or just flailing money at the wall
Zuck tried and flailed with the metaverse. That was a huge waste, but he can afford it and fortune favours the brave.
Not everyone has to make the same move at the same time.
If those don’t seem like right or good moves, I can’t imagine much will impress you in this world.
This is actually one of the hardest frontier problems. The "general purpose" assistant is one of the singular hardest technical problems with LLMs (or any kind of NLP).
I think people are easily snowed by LLMs' apparent linguistic fluency that they impute that to capability. This cannot be further from the truth.
In reality a LLM presented with a vast array of tools has extremely poor reliability, so if you want a thing that can order delivery and remember your shopping list and remind you of your flight and play music you're radically exceeding the capabilities of current models. There's a reason successful (anything that isn't demoware/vaporware) uses of agentic LLMs tend to narrow-domain use cases.
There's a reason Google hasn't done it either, and indeed nor has anyone else: neither Anthropic nor OpenAI have a general purpose assistant (defined as being able to execute an indefinite number of arbitrary tools to do things for you, as opposed to merely converse with you).
For the purposes of the exercise, let's conservatively say, maybe ~2000 tools covering ~100 major verticals of use cases. Even that may be too narrow for a true general purpose assistant, but it's at least a good start. You can slice the sub-agents however you'd like.
If you can get recall, for real user utterances (not contrived eval utterances authored by your devs and MLEs), over 70% across all the verticals/use cases/tool uses, I'd be extremely impressed. Heck, my thoughts on this won't matter - if you can get the recall for such a system over the bar you'd have cracked something nobody else has and should actively try to sell it to Google for nine figures.
There’s not just one specific solution to it either, there’s a whole class of tooling for it. And I doubt google would pay 9 figures for something that’s built on top of libraries they put out using models they developed.
As of August 1st ‘we’ (as in, I personally developed with my company, and have been paid for with real dollars which are now sitting in my bank) have a F100 using this tech in production.
As for the no true Scotsman fallacy you’re putting in front of yourself, I will let you deal with that but I would like to see how you came up with the maths.
So much attention, effort, and tooling has focused on getting llms better at writing more and more code. They can grep and curl and run scripts and iterate and build things really fast, and maybe even maintain it if given enough guardrails and direction.
But it turns out we have had a _ton_ of useful training data for models to work with for software. Not just books or docs, but examples, tests, snippets and full programs for just about any language. Show me a stackoverflow with playwright scripts or API calls (hah, as if thats possible) to build itineraries from delta, aa, united, priceline, expedia, etc, .... which is one part of one piece of the ai assistant pipe-dream.
I don't think its impossible as these tools get much smarter and more generally capable that we get decent assistants in other constrained, non-software domains, but it will take very good companies focusing on it for a long time. Much like any product that try to do these sorts of things.
Its so easy for programmers in our bubble to overlook the complexity involved in automating or even _describing_ simple tasks that humans navigate everyday via habit, learning, experience, and perception...all things that llms struggle with constantly.
This does seem like an embarrassing fail, but even Google has not completed replacing Assistant with Gemini. There have also been lost functionality (maybe temporary) in the process.
Every business has to make tradeoffs, it's just hard to imagine that any of these decisions were truly worthwhile with the benefit of hindsight. After the botched launch of Vision Pro, Apple has to prove their worth to the wider consumer market again.
Doesn't somebody (not named Nvidia) need to make a serious profit on AI before we can say that Tim Cook failed?
OpenAI and Anthropic aren't anywhere close. Meta? Google? The only one I can think of might be Microsoft but they still refuse to break out AI revenue and expenses in the earnings reports. That isn't a good sign.
I won't pretend to know exactly how the AI landscape will look in the future, but at this point it's pretty clear that there's going to be massive revenue going to the sector, and Moore's law will continue to crank.
I see what you're saying though. In particular is first generation gigs data centers might be black holes of an investment, considering in the not too distant future AI compute will be fully commoditized and 10x cheaper.
"failed to skate where the puck was headed" assumes that we know where the puck is going to be. We don't.
Everyone is skating towards that same spot while Apple is over by the blue line practicing their swizzles. They sure look like they're doomed. But large groups of people have skated to the "wrong spot" thousands of times. That's the entire point that Gretzky was making with his quote. He's not big enough, strong enough, fast enough to get in that scrum. They're all fighting it out and the puck slides away. To him. All alone.
Maybe that is Apple, maybe it's not. I mean, they're still learning to skate while everyone else is playing hockey.
The Mac is something like 30 billion in revenue per year, and 10 billion in profit.
The entire "generative AI" "industry" is struggling to reach 30 billion in revenue even with their creative accounting (my free Perplexity that comes with Revolut is somehow counted at full price, even though I never paid anything, and I'm sure Revolut doesn't pay full price), and gross profit is deep in the negative.
Instead Apple can’t even manage to implement speech to text that works in safari and can’t manage to make Siri not suck.
This seems "creepy" now, but people thought that about Google "reading" all your email too. The benefits of an ever present and aware assistant are just to great to ignore.
Apple's angle is that they are well liked and trusted (much more so than Facebook which people already think is eavesdropping on them to show ads) and will do all processing on device.
Not to mention, nobody else has achieved AGI. LLMs are not that.