Tim Cook teases AI ambitions & Apple's developmental advantages
During the earnings call, CEO Tim Cook dodged as many questions as he answered about Apple's AI plans, but did share some insights into the company's philosophies about the approach.
Apple's Tim Cook teases AI ambitions in latest earnings call
While Apple has consistently integrated AI into its devices and services, it has yet to unveil a flagship generative AI product or large language model. In statements on Thursday, Cook make it clear that Apple already uses artificial intelligence in products now.
He didn't detail where -- but that's been made clear over the years. But, for example, FaceID uses neural networks to securely recognize a user's face, even as their appearance changes over time, by processing and learning from various facial angles and features. Similarly, Apple Music employs machine learning algorithms to analyze users' listening habits and preferences to offer personalized playlists and song recommendations.
Cook and Apple have consistently maintained a policy of not commenting on unannounced products. But in the earnings call after the announcement, he emphasized ongoing extensive investments in AI technologies, promising exciting developments to be announced later in 2024.
"We believe in the transformative power and promise of AI, and we believe we have advantages that will differentiate us in this new era, including Apple's unique combination of seamless hardware, software, and services integration, groundbreaking Apple Silicon with our industry-leading neural engines, and our unwavering focus on privacy, which underpins everything we create, Cook said.
But, he wasn't done.
"As we push innovation forward, we continue to manage thoughtfully and deliberately through an uneven macroeconomic environment and remain focused on putting our users at the center of everything we do," Cook said.
The AI capabilities of the Apple Watch, as noted by Cook, are underpinned by the device's use of machine learning algorithms to enhance health monitoring and user interaction features. These capabilities include heart rate monitoring, activity tracking, and even the detection of irregular heart rhythms and falls.
All of these are functions that utilize AI to analyze data in real time and provide actionable insights directly on the user's wrist.
Additionally, details from the earnings call and previous reports hint at projects like "Ajax" and enhancements to "Safari's intelligent browsing." All of these leverage AI to refine user interactions and data processing without infringing on user privacy.
As a further security component, the Neural Engine is designed specifically to handle machine learning operations, enabling more efficient processing of AI tasks directly on the device, which enhances performance and protects user privacy. The chip allows Apple to embed AI functionalities across its product range.
The AI-driven features in Safari are intended to streamline browsing, making it faster and more intuitive, while "Ajax" is expected to offer sophisticated AI interactions securely, ensuring that user data is handled responsibly.
Apple's steady approach to AI contrasts with the rapid, sometimes preemptive product launches seen in other tech giants. The company prefers to develop its technologies behind the scenes.
Read on AppleInsider
Comments
"Apple's unique combination of seamless hardware, software"
Well, are we supposed to read 'Apple' in that? In that case he's not really saying anything at all. Or is he saying that combining hardware and software seamlessly, is unique to Apple? That would be painfully false.
Not that he had much option, though. It was more like a duty. No one can complain about that but the truth is, and Apple basically recognises this, that Apple got caught out here by an industry swing it hasn't been able to match yet.
No end of talk about neural engines and machine learning, LLM's etc will cancel out the fact that everyone has been doing the same over the last seven years.
When people talk about AI today it's all those elements and more. It is the more that Apple hasn't produced yet.
What Apple is talking up now (because WWDC is right around the corner and the iPad will come even sooner) is its very late entrance to the field. Although late is better than never.
Internally, I'm sure they're reflecting on how this situation came about (they literally went through an entire WWDC keynote trying not to utter the letters AI) but after shutting down the car project (freeing up resources) and snapping up startups etc, the 'Yikes! anxiety episode' is coming to a close even as rumours still persist of licencing technology from outside. Hence the talk of a 'new era' (which actually came to market a couple of years ago and has progressed at breakneck speed since. Up to now Apple has simply had to watch it race by.
It's probable the iPad AI news (from yesterday) will ship right out of the gate. As for the rest, I suppose it would be logical for that to roll out during the developer previews and reach users late this year. Still a way off.
Only then will the heat be off.
It will be interesting to see if they finally produce a full AI hardware/software stack too (interconnects, runtimes, inference...) or do what they've done up to now. Also how they plan (or not) to monetise it. Like Samsung, Google or in some other way.
I think they'll make a big deal out of what can be done locally although that really isn't something people really yearn for as a must have feature. But they should definitely flaunting what they've got so that's completely normal.
No doubt someone will create an AI based live counter to tally up the amount of times someone says AI at the keynote. LOL.
Although I have a suspicion that they will be trying to brand it in some way, and although I've joked about it in the past, I do think 'Apple Intelligence' will have done the rounds at high level marketing meetings.
And what of this "new market" and its "breakneck speed"? It has produced what? The two most recently hyped products Humane Pin and Rabbit R1 hare just train wrecks. Neither lived up to their hype and ended up being less capable that the current voice assistance out there.
Without a doubt there has been some cool stuff done in the generative AI space. What Open AI has done ChatGPT and Dall-E is impressive. Their upcoming video product looks impressive as well. But the user base for this stuff is still fairly small.
So, step back from ledge, take a deep breath and stop mindlessly parroting this narrative that about Apple being caught flat footed. Apple has been incorporating AI, even generative AI, in their products for years. Apple even claim that their AI features work, unlike say Humane or Rabbit. What Apple doesn't have is an Apple Chatbot... It's not the end of the world or even a setback.
By way of example:
https://www.cio.com/article/217638/a-holistic-view-at-huaweis-full-stack-all-scenario-ai-strategy.html?amp=1
Ascend (all variants)
Atlas cluster systems
Mindspore
CANN
Pangu
That is a full stack solution. Others have similar solutions.
Yes, 2018. What has Apple brought to market since that complete system debuted? That's a fair few years to do something.
Then came the pandemic. What did Apple have to offer?
Face masks?
https://fortune.com/2020/09/09/apple-special-face-mask-for-employees/
Or something more substantial?
https://e.huawei.com/id/case-studies/intelligent-computing/2020/when-ai-can-save-lives
Virus Detection
Genome Assembly
Mutation Analysis
Evolution Analysis
Remember that full stack solution?
I wouldn't expect Apple to go into those particular scenarios but the same underlying technology can be used for more CE environments and of course is being used. That Siri should be better by now is universally accepted.
Generative AI solutions have been mainstream for a couple of years now. Apple hasn't been in that. It is another reason it is in the situation it is.
Now that we are going multimodal things have really taken off. Where is Apple?
LLMs? TLMs? Again, Apple isn't pushing much.
https://www.marktechpost.com/2024/02/19/huawei-researchers-tries-to-rewrite-the-rules-with-pangu-π-pro-the-dawn-of-ultra-efficient-tiny-language-models-is-here/
Cloud?
https://www.huaweicloud.com/intl/en-us/news/20230707180809498.html
This may be from the horse's mouth but it isn't spin. Over a 100 Pangu models have already come to market.
Apple is behind. That is the sole reason they deliberately 'vetoed' the word at WWDC. That only put even more focus on their missing AI efforts.
Fast forward to the January 2024 earnings call and AI was now 'officially' back on the radar and not some kind of word that shouldn't be uttered.
Obviously, trying to pretend it wasn't an issue was now not an option. Now again at yesterday's call. And before that, the AI WWDC promo. What a turnaround but please don't push that it isn't behind because it's been 'using ML for some years' . That doesn't hack it. It didn't at that WWDC either and it's precisely why people picked up on it. In hindsight probably a mistake.
As for the user base being small. That is crazy. AI solutions are mainstream as they have permeated into all areas of everyday life. My mum uses the WhatsApp AI chatbot. I receive texts from people everyday with the footnote 'created using this or that AI solution' (in my case people use it to generate translated texts). They are also used in cloud and industrial settings. Health and Science. Weather forecasting etc. Security. Storage (for example real-time ransomware detection). Self driving networks...
Tim Cook calls it a 'new era' but it's already here and like I said moving at breakneck speed.
Hype? Sure. It's part and parcel of the industry. And you bet Apple will be right on that train at WWDC. But let's not try to imply that useful, groundbreaking solutions are already here.
I think Apple’s approach to AI, as alluded to by Tim Cook, is to avoid as much as possible getting caught up in the AI hype train or promote AI as a product in and of itself . For Apple AI/ML is technology, algorithms, logic, and operations that run under the hood and behind the curtain to make Apple’s products more attractive to consumers, easier to use, and more generally useful. Apple and Tim are trying the keep the focus on Apple products, not technology. Saying that Apple has already been using AI/ML in products for years is reassuring to those who are freaking out about the looming threat to world order that AI is being accused of by some folks, like government regulators and Elon Musk. Apple simply wants to create Apple products that customers love and purchase in great quantities. Apple’s not looking to earn an AI Merit Badge. From a customer perspective they don’t need to.
That said, Tim unfortunately does sometimes veer into the technobabble surrounding what’s currently getting a lot of industry attention. I think this is done to placate investors, journalists, and industry prognosticators who have already jumped aboard the “latest-must-have-thing” hype train and are freaking out that Apple is not throwing out some sort of AI super widget that shows off their progress towards getting the merit badge. So many others who’ve already earned their AI merit badge in the eyes of industry influencers don’t seem to care whether their AI super widgets integrate or fit into the workflow of whatever is hosting them. Apple cares. Unfortunately, for mainstream consumers, having Tim throwing out things like “Ajax” and “Safari intelligent browsing” or even “machine learning” are largely meaningless. If machine learning means their washing machine will somehow learn how to prevent mixing the whites with the darks, well that would be impressive.
Full stack development is end to end using multiple tools and technologies. Apple's Afib section most certainly qualifies as does Face ID just to name two examples. So, like I said, your claim really underscores what you don't know.
As I said in my post, you can't actually articulate the "more" that is missing, you have simply changed the subject and make demonstrably incorrect claims like Apple isn't doing generative AI, somehow you missed auto complete moving to a transform model which is ..... generative AI.
Suggesting that Apple didn't mention AI because they are behind is you trying to bend reality to fit your narrative. The realty is that Apple has a long history of using its own naming conventions for things. Using your logic Apple must be behind on Wifi because they use the term AirPort when talking about it. Apple has avoided Ai because there has historically been a certain amount of controversy surrounding it and from a marketing perspective it was easier to just say Machine Learning. If you check the record you will see that Apple mentioned Machine Learning a lot. So, no ... Apple isn't avoiding the term because they are behind. They are now directly saying AI because knobs like yourself have fabricated this narrative and they stuck having to address it.
Really, your best argument is Apple hasn't implemented an LLM in any of its products but then you follow it up with the completely brain dead statement "Apple isn't using much". The problem is demonstrably false. Apple has introduced several LLMs including one that can run on a MacBook Pro with an M2 processor. Spoiler alert, that is actually pushing LLMs as doing them on device with consumer hardware is rather challenging. What they haven't done is incorporated their LLMs in any products.
Lastly, you either didn't understand or are intentionally misrepresenting what I said when I said the user base is small. Go back and read the sentence again, I named the products that are currently driving the AI frenzy and said that their usage base is small. The consumer user base of those is in fact small. It's really strange you went from dismissing all the ways Apple has incorporated AI in day to day feature, claiming they need something like and LLM (which is basically saying "But look at open AI") to dismissing the point that the OpenAI stuff isn't used and focusing on the everyday AI stuff that Apple has actually been pushing. So which is it? Do the everyday small things count or not? Or do they only count when it works for you?
The reality is Apple has been incorporating AI in their products for the better part of a decade. Where ya'll seem to get confused is Apple simply introduces as AI as features and doesn't dwell on explaining that they are using AI when they do the announcements, they simply talk about the feature and how it is good for the customer. Imagine that? Talking about how something is useful rather rambling on about the technical points that users don't really care about.
This all remains me of the not so distant past when netbooks where a thing and a very vocal group of people were howling that Apple needed to create a netbook, Apple was behind the industry because they didn't have a netbook, netbook ... netbook .. netbook. Of course Apple didn't ever make a netbook, Apple created the iPad and netbooks are in the tech fad graveyard. Apple is doing what Apple has always done. They are looking at where technology is going, they are taking their time and being deliberate in what they do and making sure it results in products and features that are meaningful to their customers. Listening to the clowns that are currently chicken littering it up and rushing out some pile of garbage like Humane or Rabbit would absolutely be the worst thing they could do.
If you want to think an Apple Watch Afib functionality is representative of that, that's fine by me, but it is nothing of the sort.
'Auto complete' is like your Afib line. Picking out the odd capability doesn't mean much.
That's what I said earlier too and it's exactly why Apple is making such a big deal of AI this time around.
It is the same with ML and, again, like I said before. Yes, Apple has ML and has had for years. But just like everyone else and I will add. When Apple released its first ML efforts they were largely centered on image recognition when competitors were using it for far, far more. Not only that, but back then the speed of that recognition was not on par with the competition either.
Yes, things have changed (a lot) but even then Apple was behind. Once again it is behind and no, there is nothing from Apple that is currently shipping, that is on par with the generative solutions already available - now. There are even rumours of Apple licencing competing solutions.
At earnings calls, interviews and references to WWDC Apple isn't trying to set out its stall with current technologies, in the face of a couple of years of finding itself behind. No. It tried that when it deliberately avoided any mention of the letters 'AI' last time out. That just made people focus on the situation even more.
So. What is it doing? It is telling people to wait for the new iPad. It is telling people to wait for WWDC. It's not saying 'we are already covered' . 'We are competitive' . It's saying "It's coming, but wait". Not only that. It's even saying it out loud and before any official announcements. That is how important they see this. They are scrambling and that was even pointed out by someone here months ago (last year?), who apparently has contacts within Apple. Everyone was now working to get AI moving on their teams.
It isn't pointing to Autocomplete!
The 'more' is what I've already laid out:
Generative AI on par with what's out there - today.
Multimodal AI.
Full Stack AI solutions. Will we see Ascend, CANN, Mindspore equivalents? Or equivalents to Nvidia solutions any time soon?
Huge improvements in things like NLP, NLG and NLU. Yes, that obviously means Siri but far more too.
An example: Cars running HarmonyOS can differentiate between the driver and passengers and then between passengers themselves and understand follow up questions in context. Siri is basically different things on different devices (a mess) and in a test we did recently (just to see if things had improved) was still spewing out lists of web pages to manually extract answers from. We can forget about Siri digging into car innards of course, though, because Apple has opted to bail on that.
Widespread use of modern LLM's and use of TLMs.
Apple is behind.
That is probably going to change in the coming weeks (at least in terms of formal announcements) but it's been behind for a while now and every single AI reference from top management of late is basically saying 'stay tuned'. A tacit admission of the situation. At least they can actually bring themselves to utter the dreaded letters now.