Apple Intelligence, Apple Intelligence, Apple Intelligzzzzzzzzzzzzzzzz
- Thread starter Chris Law
- Start date
I question Apple giving Siri the new activation animation before actually making her any smarter. That feels like a pretty stunning miss. Why signal a change when there is no change?
I’ve been using the “Concise” function for certain emails and it’s not terrible. I feel a need to carefully review the results but it generally makes decent decisions and calms down my wordiness. Am not really using anything else so far. Things like generative AI to make “funny” stickers is pretty low on my list of priorities. As is not on the list at all.
I’ve been using the “Concise” function for certain emails and it’s not terrible. I feel a need to carefully review the results but it generally makes decent decisions and calms down my wordiness. Am not really using anything else so far. Things like generative AI to make “funny” stickers is pretty low on my list of priorities. As is not on the list at all.
What’s funnier is that if you disable Apple Intelligence, the animation reverts to the technicolor orb and type to Siri gets disabled.
I'm seeing this consistently myself. Only on MacOS though. It's weird. Some of them showing up in summaries are YEARS old, too.
The rumor going around currently is that the “real” LLM-based Siri isn’t going to be released until an iOS 19 point update sometime in the spring of 2026. I think Apple knows they’re running behind here, and updating the UI early is a quick and easy way to make themselves look like they’re on par with their competitors.
Similar type of thing with the commercials for Apple Intelligence. I saw one where they were advertising the iPhone 16, and when they mentioned “Apple Intelligence,” all they showed was the glowing border around the screen. They didn’t actually show it doing anything or even explain what it was!
I agree it’s a big mistake, though – the new UI implies that Siri itself is new, but then you find out once you use it that it isn’t any different. This makes me think when they really do improve it, it won’t be obvious (unless they change the UI again?) and people won’t notice because they’ll have already given up on it.
Kind of reminds me of iOS 7 where they overhauled the UI one year and then changed all the nuts and bolts on other years. I think by staggering it out it makes it less jarring for people and covers up if the new feature is full of bugs. Plus there's only so much Apple can really do on a year-to-year basis. In fact, I read there's already complaints that engineers are still having to work on all these major point releases for 18 instead of moving on to 19 like they normally would this time of year.
My guess is iOS 19 will appear to be somewhat lacking in features except a lot of enhancements to AI. It might not even really be advertised and 19 will be plugged as a "snow leopard" release with few user facing features. It's under the hood where it'll be a major update.
I actually think this is the new normal. For years Apple dropped all the new features at once in September and then just bug fixes until the next big drop of features. Apple has been steadily moving off that model for years now - intentionally spreading out new features across the year instead of dropping them all at once. This isn’t a new phenomenon that’s unique to iOS 18, it’s just more noticeable because the various features Apple has spread out all fall under the uniting (and heavily marketed) banner of “Apple Intelligence”.
On some level I’d actually read it as a good sign. When you have spaghetti code you need to extensively test the exact release you intend to drop because there are interrelated bugs and bug fixes cause regressions all the damn time - best to get all features done at once and test them together. The cleaner and more modular your code base is, the easier it is to just release individual features when they’re ready without worrying that updating feature A will cause a regression in code-block B. So I see Apples increasing move to incremental rollouts as a sign of a healthy codebase for iOS, and macOS.
I disabled it after a day.
Before trying to rewrite the OS to incorporate this useless feature, they need to fix notifications and AirPod switching so it's not hot garbage.
- The summaries are useless and are stuck on old notifications.
- It can't summarize my work email in Outlook, which is where I need it more.
- I hate email categories or reordering, so I disable them everywhere they appear, including GMail. I also turn off Conversation View in Outlook.
- Siri can't can't do rudimentary tasks it could before, like reminders and playing media
- The new light show is just more delay and context-switching on devices I use for more focused tasks
Before trying to rewrite the OS to incorporate this useless feature, they need to fix notifications and AirPod switching so it's not hot garbage.
I'm cautiously keeping the mail preview summaries on as I've found them accurate and useful (so far). The mail thread summary (at the top of a conversation view) has suddenly become my favorite AI feature. But why do I have to click a button to get it? And if I navigate off that thread and then back to it, the summary is gone and I have to click again to regenerate. If it was always there and dynamically updated as the thread grew, that'd be properly useful.
Agreed that Notification summaries are beyond worthless and frequently nonsensical. Instant off. I don't get the utility at all, it feels like the result of someone asking "where else can we spend some AI cycles?"
I mostly only use Siri on my phone, which is not AI-capable, so she remains as bone-stick-stone stupid as usual.
Agreed that Notification summaries are beyond worthless and frequently nonsensical. Instant off. I don't get the utility at all, it feels like the result of someone asking "where else can we spend some AI cycles?"
I mostly only use Siri on my phone, which is not AI-capable, so she remains as bone-stick-stone stupid as usual.
Agree on Mail summaries; optimizing the two-line preview is a small but worthwhile quality of life improvement.
What I find puzzling – and it's important to note that this is on the Mac – is the 'Priority Messages' segment that occasionally appears atop the message list. Unless I'm holding it wrong, clicking the items within this segment does absolutely nothing.
What I find puzzling – and it's important to note that this is on the Mac – is the 'Priority Messages' segment that occasionally appears atop the message list. Unless I'm holding it wrong, clicking the items within this segment does absolutely nothing.
Yeah, I don't get that either. I think it would be more useful for someone who receives many, many emails and doesn't keep their inboxes clean (which, of course, means they're an unhorsed savage as well). So it would be a decent indicator of an important or vital message among many, but if that was the target situation... then absolutely, clicking the notice should send you to the relevant text. So...
Yep. UX wizdoms: if your users know ducks, and you render something like a duck, better make sure it quacks like a duck.
So you’ve met me then.
(Sadly, Playground doesn’t recognize the concept in a meaningful — or entertaining — way.
I’ve just spent a few minutes futzing around with the image playground. First impressions are that it seems incredibly limited. The text input recognizes almost nothing; it gave me unpleasant flashbacks to the Zork parser. Plus it made me look like Walton Goggins with a hint of Henry Winkler, which I found to be mildly upsetting.
Apparently I look older than I feel because every damn image of me the playground creates is like “who the fuck is that old guy supposed to be? Me? GTFO!”
Then I look at myself in the mirror and… damn…
I guess I like it because it genuinely created a YOUNGER image of me, that others still instantly recognize as me. So in my book - that's a winner.
This might have been fixed in 15.2 – I was having the same problem for weeks with Reminders notifications specifically, but I just noticed today they were accurate/up to date.
That’s weird, I get Outlook notification summaries just like I do with Apple Mail. I didn’t do anything in particular to set it up.
One other thing though with notification summaries that Apple really needs to address: it doesn’t account for spam. If a spam iMessage or email makes it through, it summarizes it and (because of the nature of spam) frequently marks it as “maybe important.” I get that spam filtering and notification summaries are two different features, but having the computer tell you that a phishing text is an important message is dangerous and worse than not doing anything.
Well, this is bleak: BBC complains to Apple over misleading shooting headline. I don't think we need Apple Intelligence just making up headlines and attributing them to a reputable news operation.
Good rundown of Apple’s image gen restrictions: What You Can and Can't Make With iOS 18.2's Genmoji Feature
is your instance of Outlook on your iPhone behind an SSO?
That may be blocking access to my email from the process.
I gave it a few weeks and disabled it once I saw Image Playground. Everything 'Apple Intelligence' has been terrible, no matter how much they emphasize beta.
I like where Google is going with their NotebookLM (which just got even more impressive features today) and ChatGPT still remains the go-to for general purpose stuff.
I don't think Apple will even come close for at least a few years.
Of course they won't; part of the quality acceleration ramp of the other services is a wanton disregard for privacy or ethics, and just feeding the machine any and all scraps they can find.
Apple almost certainly syphoned up massive amounts of copyright data in their own training. They are all drinking from the same well, some are just more brazen about it than others.
And they all outsource their training, labeling, and guardrails through a labyrinth of agencies. Apple might have their labeling system for hardware sourcing and keeping children out of their manufacturing lines, but they are conspicuously silent about how they use third-parties as a 'moral crumple zone' for training AI (just like all of the other major tech companies).
They might be less unethical about it, but there's no way they limited their data to common domain material only. Honestly, copyright law is not prepared (or sufficient) for generative AI.
The other companies rushed to grab as much as they could because they knew that the door would eventually slam shut on that fountain of data and they wanted it locked in before that happened. The former chief scientist of OpenAI, Ilya Sutskever, even recently said:
And they all outsource their training, labeling, and guardrails through a labyrinth of agencies. Apple might have their labeling system for hardware sourcing and keeping children out of their manufacturing lines, but they are conspicuously silent about how they use third-parties as a 'moral crumple zone' for training AI (just like all of the other major tech companies).
They might be less unethical about it, but there's no way they limited their data to common domain material only. Honestly, copyright law is not prepared (or sufficient) for generative AI.
The other companies rushed to grab as much as they could because they knew that the door would eventually slam shut on that fountain of data and they wanted it locked in before that happened. The former chief scientist of OpenAI, Ilya Sutskever, even recently said:
Last edited:
It’s fun if you’re not trying to flog it. This “kitten in a space suit” is pretty cute. (I don’t have a device capable yet, so I did this at the Apple Store today.)
Edit: Spoilered for big now that I'm on a computer. I didn't realize how big it was on my phone.
Edit: Spoilered for big now that I'm on a computer. I didn't realize how big it was on my phone.
Yes, I'm being picky, but hears and paws outside the suit?
It is cute, and I'm a dog person.
Yeah, it seemed very limited in the brief time I played around with it. The characters it creates look almost nothing like the people they’re based on; it just picks up on basic features (hair color, facial hair, glasses, etc.) and then creates a generic-looking character with those same features.
The text input doesn’t seem to recognize anything beyond basic terms, and gets confused even then; for example, I wrote “[me] driving a green car down the highway” and the images it created all had me wearing a green shirt, rather than the car being green.
Image playground stays mostly in a “cartoon” image space and in that context ears and paws sticking out are not a surprise. I don’t expect any realistic images from it yet. At this stage it is more of a silly toy than a serious tool and that’s fine.
Seems like it would be an easy fix for the notification API to include a means to disable summaries. It's not like Apple's 1st party notifications always get summarized.
I get it. That said, it's perfectly cromulent for a throwaway image to post in a group chat with the kids.
That's odd. I tried with both of my kids right away, and they were both very recognizable as themselves. It did a passable job with me.
I am kinda glad I didn’t get a new phone this year! I will continue to play with it on my iPad I guess, but until I start hearing folks rave about how useful it is for stuff using the camera (for out and about use), or that Siri isn’t so … Siri. Also, the sphere (at least on Apple Vision Pro) is super crazy neat. Making the screen light up is something they coulda held back until Siri was “new and improved”, so I don’t develop the same revulsion to the new animation that I currently have for the sphere.
Also, how about a little apple intelligence regarding AirPod routing and Siri routing. Can’t there just be one Siri to rule them all, not just every device in earshot and (without fail) the one that answers will be the one that cannot do what I ask.
Also, how about a little apple intelligence regarding AirPod routing and Siri routing. Can’t there just be one Siri to rule them all, not just every device in earshot and (without fail) the one that answers will be the one that cannot do what I ask.
An even easier fix would've been to anticipate something like this might happen and think better of doing it in the first place. We've known for while that "AI" is a fabulation machine that approximates reality often enough to fool certain kinds of primate.
Had a play with image playground (after getting confused by the app in the appstore with the exact same name). As thumbnail images they're okay but limited, tends to give very minor variations on a theme and ignore a decent bit of what your requesting. As soon as you click on the image you can see lots of defects.
Honestly it feels more like playing with a well tuned SD1.5 rather than something modern like FLUX, FLUX seems to kick it's ass from my experience. It also appears to have a tiny prompt/context window for your requests, and restricted to cartoon style images.
Honestly it feels more like playing with a well tuned SD1.5 rather than something modern like FLUX, FLUX seems to kick it's ass from my experience. It also appears to have a tiny prompt/context window for your requests, and restricted to cartoon style images.
Welp, I’ve had a few days to futz around with the AI bells and whistles in 18.2. My reactions ranged from indifferent to visceral hatred, mostly on the latter end of the spectrum. I’m not quite ready to call Apple Intelligence a disastrous waste of everyone’s time, but man does it have a long way to go before it becomes anything close to something that I want. I’ve turned it off, and will check in after another revision or three.
My best hope is that in a few years we can look back on this like Apple Maps 1.0, and marvel at how far they’ve come from such inauspicious beginnings.
My best hope is that in a few years we can look back on this like Apple Maps 1.0, and marvel at how far they’ve come from such inauspicious beginnings.
It's Memoji or FaceTime filters. Great for a demo and then completely ignored. I don't think anyone will seriously (siri-ously?) engage with it for the foreseeable future.
Hm, yeah I tried with my kids too and got better results. Maybe they have more distinctive features, or maybe the art style lends itself better to kids?
When I try it on myself and my wife, the characters don’t look like us, they just look like people that have the same features. Does that make sense? Like if you told an artist to sketch a portrait of your friend, but you just described them verbally and didn’t share a photo for reference.
But whatever. As a free tool to play around with for fun, it’s fine. It’s just not very impressive.