Apple squandered the Holy Grail

I think it is concerning that every single Apple Intelligence feature they&#x27;ve shipped thus far has been not just mediocre; but bad. Being last to the party is a very normal Apple thing; quality and Doing The Right Thing takes time. Announcing something then taking months to ship it is very not-Apple, but it has happened a few times. That thing they finally ship being bad is, geeze, horribly un-Apple.<p>One of the few examples I can think of however is Apple Maps. And it did get better; a lot better, some say better than Google Maps nowadays. So I generally do have hope for Apple Intelligence. At the end of the day, there are some disparate competing utilities in this class on the Samsung and Google phones, but no one is shipping something that is obviously game-changing and in first place; they all kinda suck, they&#x27;re all tech demos, and it&#x27;ll inevitably take many years to get this technology honed in to something that is truly useful to consumers.

The &quot;bicycle for the mind&quot; goal, and the Steve Jobs quote that inspired it, is really just another restatement of McLuhan&#x27;s idea of media as being extensions of man. The bicycle (or wheel, more generally) is an extension of the legs, a phone is an extension of your voice, etc.<p>The problem with interpreting AI through that lens is that AI, as it is being used here, <i>is not an extension</i> of your mind. Plenty of other things are (organizers for example), but AI does not extend your thoughts. It <i>replaces</i> them. Its notification summary feature does not improve your ability to quickly digest lots of notification information, it replaces it with its own attempt, which, not being your own judgment, can and does easily err.<p>There are some uses of AI that do act more like a McLuhanesque medium. Some copilot applications, in which suggestions are presented that a user accepts and refines them, are examples of this. But a lot of the uses of both image generation and LLM tools serve to limit what your mind does rather than expand it.

This post makes the point that the foundations of Apple Intelligence are really well designed. I think anytime you make the right underlying technology choices, there is always hope for the product.<p>It&#x27;s also worth noting that Apple traditionally is not a first mover and looks for &quot;inspiration&quot; from smaller competitors. In this case, there is no comp to reference. There is no startup mobile OS innovating in integrated AI. That, and the supposedly rushed timetable, probably explains a lot.

&quot;Clean Up is best explained by this famous photo editing example . . . This tool allows you to capture a moment in time as you wish it happened, not as it actually happened.&quot;<p>FALSE. Apple defines a photo as a record of something that actually happened. iPhones take photos. They doen&#x27;t auto-swap a high-res moon in for the real one like Samsung phones do.<p>Clean Up (like crop) is just an editing feature, manually applied after a photo already exists, and using it effectively changes the image from a photo into an &quot;edited image&quot;, the same way using Photoshop does.<p>Definitions of What a Photo Is:<p>Apple - &quot;Here’s our view of what a photograph is. The way we like to think of it is that it’s a personal celebration of something that really, actually happened. Whether that’s a simple thing like a fancy cup of coffee that’s got some cool design on it, all the way through to my kid’s first steps, or my parents’ last breath, It’s something that really happened. It’s something that is a marker in my life, and it’s something that deserves to be celebrated.&quot; - John McCormack, VP of Camera Software Engineering @ Apple<p>Samsung - &quot;Actually, there is no such thing as a real picture. As soon as you have sensors to capture something, you reproduce [what you’re seeing], and it doesn’t mean anything. There is no real picture. You can try to define a real picture by saying, ‘I took that picture’, but if you used AI to optimize the zoom, the autofocus, the scene — is it real? Or is it all filters? There is no real picture, full stop.&quot; - Patrick Chomet, Executive VP of Customer Experience @ Samsung<p>Google - &quot;It’s about what you’re remembering,” he says. “When you define a memory as that there is a fallibility to it: You could have a true and perfect representation of a moment that felt completely fake and completely wrong. What some of these edits do is help you create the moment that is the way you remember it, that’s authentic to your memory and to the greater context, but maybe isn’t authentic to a particular millisecond.&quot; - Isaac Reynolds, Product Manager for Pixel Cameras @ Google<p>Definitions via <a href="https:&#x2F;&#x2F;www.theverge.com&#x2F;2024&#x2F;9&#x2F;23&#x2F;24252231&#x2F;lets-compare-apple-google-and-samsungs-definitions-of-a-photo" rel="nofollow">https:&#x2F;&#x2F;www.theverge.com&#x2F;2024&#x2F;9&#x2F;23&#x2F;24252231&#x2F;lets-compare-app...</a>

&gt; Many companies want to make computers that you can use to do computer things. Apple makes tools that you use as an extension of your body in order to do creative things. They don&#x27;t just sell computers, they sell something that helps enable you to create things that just so happen to be computers.<p>An Apple marketing executive is smiling somewhere. Brainwashed another one!

They failed IMO as soon as they started marketing this stuff as AI. Nobody _actually_ cares about that. They just want new and exciting device capabilities. The fact that they haven’t done anything compelling enough to detach it from the AI moniker means they don’t have anything. Yet. The overall foundation is sound. Maybe in a generation or two we will have a truly compelling use case. Part of me doubts it, though, because Apple isn’t the company they used to be when it comes to innovating.

Sometimes I wish I could check an alternative universe where Apple refused to acknowledge AI slop and just doubled down on privacy and protections of the user. It almost feels like they could have just ignored the trend and let everyone else burn money until the hype dies down.

&gt; The core of why ChatGPT works as a product isn&#x27;t the AI. It&#x27;s the experience of each word being typed one at a time by the AI and saving your conversations with the AI for later.<p>Really loved this article overall, but I have to super-disagree here. The core of ChatGPT is you can have a conversation with a computer program.<p>Take away saving history and you can still have a conversation with a computer program (see ephemeral chats).<p>Take away typing one word at a time and you still have a conversation with a computer program (see non-streaming API &#x2F; batch API).<p>But major props for writing this live on a twitch stream, benefit of the doubt there my friend.

The “play the podcast my wife sent me the other day” example is interesting to me. That shouldn’t be difficult to do <i>without</i> AI. Yeah asking a thing is always gonna be quicker (provided it works), but a well designed app should make that possible within like ten seconds.<p>I can’t help but wonder if the reason “agentic” systems seem so appealing to people is because as an industry we’ve spent the past fifteen years making software harder to use.

&gt; Word processors like MacWrite absolutely transformed the ways that everyone used computers.<p>MacWrite was released 5 years after WordPerfect, which itself is predated by WordStar. I don&#x27;t get why Apple fans have this obsession with pretending Apple invents these things.<p>Apple refines what others have attempted before, that&#x27;s what they&#x27;re good at. Part of the reason people are disappointed with Apple these days is because of this fantasy image of Apple as an inventor.

I am hoping the foundation they built will lead to greater things. So far it definitely has fallen flat compared to their demos. All I wanted was a way to talk&#x2F;text to siri in a natural way to get things done. It&#x27;s better but far from perfect. I want to be able to easily create calendar events and interact with other native iOS APIs.

I don&#x27;t get the &quot;Math Notes&quot; example. Why does this need AI? Isn&#x27;t this just Calca[1] but with extra steps?<p>[1] <a href="https:&#x2F;&#x2F;apps.apple.com&#x2F;us&#x2F;app&#x2F;calca&#x2F;id635757879">https:&#x2F;&#x2F;apps.apple.com&#x2F;us&#x2F;app&#x2F;calca&#x2F;id635757879</a>

While I agree with the post&#x27;s sentiment, its assessment of Apple Intelligence overlooks its incomplete rollout, with most of the meaningful, step-change features (e.g., Siri’s contextual awareness) scheduled for release in 2025 at the earliest.<p>In my view, Private Cloud Compute and Apple Intelligence, together with the ubiquity of Apple devices, position Apple as a leading candidate to realize the widespread, AI-enabled transformation of personal computing discussed in the post — with tasks requiring less energy and cognitive load than they do today for the general consumer.

I could pick on quite of few nits here but I&#x27;m going to focus on one in particular that I&#x27;m very familiar with as a photographer and mass media studies student.<p>&gt; I want the data coming off of the sensor to be the data that makes up the image. I want to avoid as much processing as possible and I want the photo to be a reflection of reality as it is, not reality as it should have been. Sure, sometimes I&#x27;ll do some color correction or cropping in post, but that doesn&#x27;t change the content of the image, only its presentation.<p>First nit: the iPhone camera, and all digital cameras, are deeply influenced by computational photography techniques. What this means is that you essentially <i>never</i> get the raw pixel values, although there are exceptions. The image you get is <i>already</i> significantly manipulated.<p>Second nit: color correction, color in general, dynamic range, focus, depth of field, and more are all manipulations made by default, even long before digital cameras when film was king. There is no &quot;correct&quot; image version of what our eyes see, there is only pleasing to the photographer and the audience.<p>An example: the negative for Ansel Adams&#x27; well known &quot;Moonrise Over Hernandez , New Mexico&quot; looks like, at first glance, something a professional would trash for lacking detail.<p>Here&#x27;s contact print vs the version most of use will probably recognize: <a href="https:&#x2F;&#x2F;images.squarespace-cdn.com&#x2F;content&#x2F;v1&#x2F;5f5fe5ca8d6a356d4136ec94&#x2F;1620499484148-PJP8X728J79GWLZGEVYD&#x2F;beforeafterMoonOverHernandez.jpeg?format=2500w" rel="nofollow">https:&#x2F;&#x2F;images.squarespace-cdn.com&#x2F;content&#x2F;v1&#x2F;5f5fe5ca8d6a35...</a><p>Here are four different versions Adams printed over the course of 3 decades: <a href="https:&#x2F;&#x2F;images.squarespace-cdn.com&#x2F;content&#x2F;v1&#x2F;5f5fe5ca8d6a356d4136ec94&#x2F;1620499552352-R7VDU80OD5YWT2QWMN3F&#x2F;moonrise-printings.jpeg?format=2500w" rel="nofollow">https:&#x2F;&#x2F;images.squarespace-cdn.com&#x2F;content&#x2F;v1&#x2F;5f5fe5ca8d6a35...</a><p>I will mention, but won&#x27;t even get into a topic that will surely bait HN commentors: Kodak designed and standardized its color film to represent Caucasian skin tones. It wasn&#x27;t until chocolate and furniture makers complained that everything looked like the same gross mud in their expensively-produced product catalogs that Kodak took a look at rendering dark brown&#x2F;red&#x2F;yellow tones more pleasingly. Notice I said &quot;more pleasingly&quot;, <i>not</i> &quot;correctly&quot;.

There is also the possibility that LLM models (which is what Apple Intelligence is leveraging mainly) are overblown and honestly a local minima for AI.

Y&#x27;all know I&#x27;m a fast, if error prone, writer. I still enjoy using AI writing assistants to help me with the occasional phrase that&#x27;s awkward, grammar detail (&quot;it&#x27;s lower g in &#x27;god&#x27; if I am talking about Thor or Huxian right?&quot;) and choice of words (&quot;I need a word for agriculture that starts with C...&quot;)<p>LLMs make different mistakes than I do so I&#x27;ve thought about using one as a copy editor but I&#x27;ve had terrible experiences with copy editors: I&#x27;ve hired more than one when I was writing marketing copy who injected more errors than they fixed. (A friend of mine wrote an article for <i>The New York Times</i> that got terribly mangled and barely made sense after the editors made it read like an <i>NYT</i> article.)

I normally really like this blog, but what’s all this noise about? They’ve done the hard part of lining their entire foundation up, now all they have to do is build the applications on top, and let others do so as well.<p>In the meantime, they’re shipping the best non-workstation computers, by far, to run models locally.<p>They don’t have to be the ones to implement all of this themselves, you can install ollama right now, and BoltAI can integrate those models into other parts of the OS. And Apple will watch, and Sherlock the best parts of what others do into the OS, and sell gobs of machines.<p>They haven’t squandered anything, the foundations are still there.

It&#x27;s a bit funny his favourite &quot;Apple Intelligence feature&quot; is something that wouldn&#x27;t surprise me if it doesn&#x27;t even invoke the actual model at all under the hood.<p>Parsing text for variables when it sees an equals sign and running basic calculations on them? I feel that could have been a novel feature 30 years ago.

I was an early user of Macs; the first computer I owned was a Mac Plus. I didn&#x27;t own a PC until the 1990s.<p>Most of the intro to this is credulous hooey. Macs weren&#x27;t &quot;bicycles for the mind&quot; in some magic way that was different from PCs. What the early ones had was 1) a better and much more standardized interface, and 2) task switching that worked.<p>As for the AI tools, image generation might occasionally be useful for a D&amp;D game, but otherwise nothing on offer at the moment has much value. And the value of image generation (for me) is pretty small.

They just built a trusted&#x2F;secure backend to push compute to and it luckily coincided with the AI craze. They just packaged their backend as apple intelligence and exploited the situation. It doesnt look like they have anything worthwhile to showcase that backend though. They will get there eventually, this is apple after all

&gt; Hell, the iPhone is a fully capable cinema camera these days<p>No, it&#x27;s not. The sensor in an iPhone is AI&#x2F;ML&#x27;d up the ass to hide all the noise because it has 1µm sensor wells.<p>A Panasonic video-oriented mirrorless micro 4&#x2F;3rds (so not even anywhere near 35mm) like the GH5 is 3-4x that.<p>A Sony Alpha 7 III? <i>six</i> times the sensor well size.<p>I don&#x27;t care how many megabits of video bandwidth you throw at it or how fancy you think &quot;raw&quot; shooting is, or how fancy your sensor technology is; nobody these days has anything that is even close to 2x better than anyone else. The top sensor from all the major players are pushing the limits of physics, and have been for a long time.<p>No amount of AI&#x2F;ML shit will give you depth of field and bokeh that looks as nice as a big sensor and a fast lens with nice shutter leaf shape.

It&#x27;s astonishing to see Apple settle for DALL-E 3 (or worse?! these remind me of the Bing samples) for the image generator part. Hasn&#x27;t the incredible extent of mode-collapse and the horrible DALL-E 3 style become universally known and disliked yet?

&gt; If you have any modicum of site reliability experience, this seems like an unsatisfiable set of constraints. It seems literally impossible, yet here they are claiming that they have done it.<p>His first instinct was right. It seems impossible, because it is. Unless I can run the entirety of the &quot;Private Cloud Compute&quot; on my own hardware in my own firewalled network, I 100% believe that the pipeline is compromised; our data is siphoned off and sold off to advertisers, especially now that they know they can do it and get less than slap on the wrist: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=42578929">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=42578929</a>

&lt;glances at recent story reconfirming Apple as the most highly valued private company the world has ever known&gt;<p>This is sardonic, as yes, Apple could have chosen different monopolies than it currently has, at different points, and had a different (maybe not better?) trajectory, some of of us are old enough to remember the antipathy towards Microsoft when to the Office suite it added default Explorer,<p>But also, maybe our system fundamentally rewards the &quot;wrong&quot; things if one&#x27;s definition of &quot;right&quot; includes things like innovation. Or maybe, the welfare of the commons and the common good.

Mac’s being able to export PDF’s for free was a huge deal back in the acrobat days :-P<p>I’m hoping with what they’ve built in infrastructure and custom chips is a step towards making personal LLM also highly available to non-technical people. I think this is where Apple has always shined - making things not just better, but accessible and grokable for normal people.

What&#x27;s interesting is how I opened Safari&#x27;s reader mode to digest this 5000+ word polemic, and then noticed for the first time a new option to summarize its contents. A few seconds later I had a clear idea of what the author&#x27;s thesis is, without being under the false impression that it had conveyed its finer points.

The Apple bicycle is the Apple user. They know how to ride it well. And how to take the Money.<p>Generative AI is a dud. ML has a lot of applications. I use my smartphone for calls, chat and banking apps. I use my iPad only for drawing, mail and some games.<p>For everything else, I have computers. With real OS.

I was giving this my attention until the author included a long quote from Steve Jobs <i>from the author&#x27;s own dream</i><p>Sorry, what?<p>Apart from the level of dream-detail recalled being highly dubious, quoting your own hallucination of Steve Jobs to help with your argument about generative AI being useless (and missing the irony) is downright weird.<p>Also Math notes is basically the same thing search engines have been able to do for over a decade now. Enter a sum, get an answer.

So far it feels very unfinished, but having followed Apple for a long time I&#x27;ve seen many products launched and iterate over time. Maps, for instance, had a fairly disastrous early period but eventually became my preferred navigation app.

I’m withholding judgement on Apple Intelligence until iOS 18.4 is shipped in May. That is when they plan to release a revamped Siri with better contextual and personalized responses. For instance, AI&#x2F;Siri will be aware of what is currently on the screen when responding and also integrate personal data across apps.<p>Ultimately Apple’s strategy of a privacy focused AI will be a winner for a consumer device with access to sensitive personal information. It’s a question of whether they can pull it off technically.

I&#x27;m going to admit that I just skimmed past 90% of this article. Being dismissive of AI is currently easy content, so there&#x27;s too much noise in the space.<p>Having said that, I actually paid attention to the image playground criticism. Image playground is <i>literally</i> a playground. It is meant to make fun, low-effort images for friends and family, largely for social type interactions.<p>&quot;It uses a placid corporate artstyle and communicates nothing.&quot; It&#x27;s a hot taco holding a beer. What is it SUPPOSED to communicate? Looks like a pretty great image to me. But of course this piece was leading into the anti- angle, so suddenly it&#x27;s &quot;horrifying&quot;. I guess I didn&#x27;t get the special training to understand what was wrong with a clearly lighthearted, fun image.<p>Similar asinine, overly-jaded complaints about the cartoonish, memoji style portrait generation. I think the image is actually pretty hilarious. Actually used image playground to make my social media image, and I care not what this guy thinks about it, or that it is &quot;soulless&quot; (as if a cartoonish representation is supposed to be soulful?)

It feels very apparent to me that Apple was caught completely flat footed by ChatGPT’s release. As a result, they were very far behind and have been unable to catch up.<p>What they released as Apple Intelligence wasn’t a well-planned, cohesive product as much as the only thing they could possibly do, given the timelines they were up against. Maybe they’ll catch up, but they’re definitely behind and it’s a shocking thing to behold.

The biggest problem I have with Apple Intelligence is battery life. Since Apple has no software chops in building LLM models, I expect them to throw hardware solutions for the battery life problem.<p>But the demands of intelligence and the general trajectory means no amount of hardware - storage, RAM or battery size would be enough to generate the high fidelity experiences or solutions that fans and customers have come to expect from the company.

This is Apple being smart and hedging their bets.<p>They are dedicating “AI” hardware to their current devices, and are building a framework for many different futures. In regards to software, they are first building the safety, then MVP’s to provide APIs to their chips.<p>So yah, maybe they aren’t holding the holy grail at the moment. At least we won’t be drinking from lead goblets .

Their problem is that they are just behind now, because they released too early. They haven’t finished iOS 18 yet. They are still working on it, which likely means that iOS 19 isn’t going to get much attention either, because they should be working on iOS 19 now, not still developing iOS 18.<p>They have set themselves up for a loser in the next year or two, because they can’t double their resources to catch back up to a normal release schedule.

Stepping back the way I see it is it’s a cultural thing. Apple loves to own the whole stack and even though circumstances forced them to use the hot new thing because it’s useful, Apple as an organization doesn’t really wanna do that. It wants to have its own stuff, and so it could never really make something succeed that wasn’t its own and that’s why it failed. So far.

Hah I’ve been having the same issue as the author with those scammy “package delay” texts getting summarized in my notifications.<p>Didn’t realize how widespread that type of spam was until now. Why hasn’t someone implemented better spam detection at Apple like we have for email? It would be nice if they could classify texts as spam, promotions, etc and organize them the way Gmail does.

&gt; Why Apple Intelligence failed even though everything it&#x27;s built upon is nearly perfect<p>Because that’s a laughably false premise.

This article makes great points, but the outcome seems to be “in progress” and not final. They have the tech stack and the right philosophy, they just botched the execution out of the gate. This is the same company that didn’t think 3rd party apps on the iPhone would be a thing. They managed to course correct.

&quot;Apple Intelligence failed&quot;<p>&quot;Apple Intelligence&quot; is less than a year old. Give it some time, for chripe&#x27;s sakes.

A somewhat tangential observation:<p>Apple has done such a good job with marketing that my 10 year old thinks that AI stands for Apple Intelligence.<p>We live in the Bay Area so she&#x27;s seen a bunch of billboards with that. I have to constantly remind her that that is not what it means in most cases.

Math notes is a Fortan repl. It’s only impressive to someone who has never seen a calculator.

It&#x27;s not just about the man on a bicycle, it&#x27;s a man on a bicycle on a flat road.

Apple has a history of releasing early proofs of product concept or basic table stakes products built with a high enough quality they feel like they’re “done,” but then proceed to methodically iterate on them for years to decades until they fade into the background. It’s hard to imagine the iPhone essentially started as a click wheel iPod - comparing the two is night and day in terms of capabilities, function, and form. They take weird side ways jaunts - but generally shift back into a path that is sensible. I was interested to see what they did with Apple intelligence, but assumed it would be establishing of a basic set of capabilities, the effective proposals for APIs, and the seeding of product discussions with their customers over a long time. People seem to think a few years into the current cycle of AI technology we are seeing the final fruits rather than seeing the infancy - for those who develop these sorts of tools it feels very much like iPhone gaming felt in 2009. At some point over the decade we will hit the nadir, then descend into total enshitification (and yes those who think AI has already reached peak enshitification you are totally wrong). Along the way though we will see a lot more truly stunning advances towards the final arrival at pervasive exploitation.

&gt; Except for Math Notes. Holy crap. I love Math Notes so much. I wish other note-taking apps had it. It&#x27;s easily the best feature they&#x27;ve ever come up with.<p>The Cursor editor literally has this.

&quot;Apple Intelligence failed?&quot; The far-reaching project that just had its initial release like 60 days ago? Why would anyone read beyond a first sentence like that?

It does seem a little sloppy, but the actually interesting part of Apple Intelligence isn&#x27;t out yet so I&#x27;d withhold judgement, even on the initial release.

I think part of this issue is that people expect a lot from Apple. They exposed the technology to many people who aren&#x27;t early adopter types, but more &quot;mainstream&quot; types. With an entirely new product and brand, the tolerance for bugs is higher, but here the expectation is that of other Apple products.<p>In the end, even if the features aren&#x27;t perfect, they still raise the bar for competitors, so Apple is less in danger of being disrupted.<p>Also, there is plenty of AI driven features that people do not talk about, but those &quot;just work&quot; so you don&#x27;t see them as well.

It&#x27;s what Apple does. The first version of a product often sucks somehow, but is generally usable. They&#x27;re masters at iterating.

Well - while Apple has made rough starts in the past (Maps on iOS devices comes to mind) - they do have a solid track record.<p>Having used &quot;smart devices&quot; since the Apple Newton 2.0 days, followed by Windows Mobile, a very brief Android excursion (Motorola Milestone - early enough Android that I was often frustrated trying to copy&#x2F;paste text between apps), then another side-pivot into Windows Phone for awhile (mainly because the development was incredibly easy - and Microsoft gave me a free one), I have been in the iOS mobile phone ecosystem ever since the iPhone 6.<p>And - the software has gotten increasingly better over time - I wouldn&#x27;t have (for me) alot of content&#x2F;subscribers on TikTok, if iMovie on my phone did not exist - attempting to edit videos using OpenShot was taking forever (While I have Davinci Resolve installed, it seems &quot;daunting&quot; for someone who doesn&#x27;t want to be a professional videographer&#x2F;editor).<p>But then I tried iMovie &quot;Magic Movie&quot; on my phone and ... &quot;it just works&quot;. Still not great for long-form YouTube style content, but for quick things, slice-of-life videos - it does the job rather well.<p>... I expect that Apple will improve these AI offerings dramatically over the next couple of years as people upgrade their devices.

Apple also squandered maps ... until they didn&#x27;t.

The image playground app is an embarrassment

This is very funny. The picture of Stalin and Yezhov was exactly what was going through my mind as an Apple Store sales guy explained the photo clean up feature to me. It felt rude to bring it up though.

&gt; If this OS were shipped to consumers, you would have a nearly unhackable system that would make it basically impossible to tinker with.<p>I mean you can hack it the same way you would hack any other Darwin platform

&gt;Then they casually dropped the holy grail of trusted compute...<p>By which I think he means the AI stuff runs on your machine rather than the cloud. For me that&#x27;s not a holy grail at all or even something I&#x27;m terribly interested in. I downloaded Apple AI on the macbook, found it quite meh and am now seeing if there is a way to remove it as it uses quite a lot of GB or memory. I can see for someone wanting to use LLMs on confidential corporate data that would be important but that&#x27;s a specialist use case that I don&#x27;t think Apple Intelligence is particually good for.

Imagine how much better you would be at typing if you didn&#x27;t have a backspace key.

This is why I switched to Guix lol

I find it super interesting that Apple is x-raying the PCBs for their compute nodes. Guess they took Supermicro inserting malicious devices into their servers very seriously.

Apple shipped “Apple Intelligence” before Apple even invented the term.<p>Before the AI craze you could search Photos on iOS based on content and metadata. You could lift subjects off photos with a long tap and copy them, recognize faces and make montages based on inferred relationship with them. And all of this is done on-prem, on your device.<p>These are very subtle, nice features. Apple had to put a name on all these otherwise there would be no marketing material.

&gt; They sell bicycles for the mind.<p>Given the lack of a single good supporting example (what, did PCs have no word processors that reacted to backspace keys?) it seems like these are fantasy bicycles...<p>And since no evidence is needed to believe, you can of course believe in Intelligence that can act better than your brain (what are &quot;those pics from San Francisco&quot;, you&#x27;ve snapped a hundred there, which 5 would you like to post?)<p>And yet the disillusionment comes a bit faster than expected, why not give Apple a few more decades to iron out some kinks on such a revolutionary fantasy path?

It’s ok to just say you don’t like Apple Intelligence.

I&#x27;m still having to look for thunderbolt cable all around my house to charge that one freaking device that is iphone when everyone else switched 10 years ago to USB C.<p>How is apple intelligence going to help me with that ?

Apple was always going to fail this, and even more so going forward.<p>LLM are built on data, and copious amounts of it. Apple has been on a decade long marketing campaign to make data radio active. It has now permeated the culture so much so that, Apple CANNOT build a proprietary, world-class AI product without compromising on their outspoken positions.<p>It is a losing battle because the more apple wants to do it, the users are gonna punish them and meanwhile, other companies (ChatGPT, anthropic) are gonna extract maximum value.

Apple has started going downhill from the time they released Vision Pro.<p>So that means all of the big tech companies are going down: Facebook, Google, Apple.<p>Only Microsoft remains strong though for how much longer remains to be seen.<p>A great time for startups.

I feel like they have a plan, get the backend up to scratch to appease the tech people and they&#x27;re leaving the ways to interact with it purposely vague and incomplete for those non-tech folk.<p>They don&#x27;t want to scare any part of their audience away from future uses of Apple intelligence. Their audience is tech and non-tech folk alike.<p>If the tech folk say it&#x27;s safe and the non-tech folk get comfortable with the basic AI features then they&#x27;re onto a winner.<p>How many people&#x27;s parents&#x2F;grand parents have iPhones because they&#x27;re simpler for them to understand who are also scared or don&#x27;t understand this &#x27;AI thing&#x27;. I think Apple have been quite savvy in introducing it slowly and are probably watching the metrics like a hawk!<p>I suspect image playground is so creepy in an attempt to mark the images as clearly AI generated when they get posted to social media?