ChatGPT is treated like a search engine for Siri, and generates text and images for apps.
See full article...
See full article...
Apple integrates ChatGPT into Siri, iOS, and Mac OS
- Thread starter JournalBot
- Start date
Upvote
224
(278
/
-54)
Just days ago you could see comment after comment stating "Well, at least Apple is being cautious with AI.", "Well of course Apple is taking a 'wait-and-see approach', they're the adults in the room", etc.
This does not bode well.
This does not bode well.
Upvote
73
(173
/
-100)
I want a single master switch to permanently disable this. Just one toggle in the settings, and if ChatGPT is set to "off" there, then it will never go to that site. I'd rather have no answer than an AI generated one. I really can do without a bullshit generator in my pocket.
Upvote
190
(240
/
-50)
For those alarmed by this, I'm not seeing where Siri offering up an option to enlist ChatGPT in your results has any downside. Just don't don't click on the permission? Unless I guess just seeing such a prompt is an enraging violation of some imagined line in the sand?
Upvote
83
(178
/
-95)
One of the liveblog mentioned Sam Altman was in attendance and my heart sank.
Then the actual preso happened and all of the ML features seemed really reasonable and like good uses of ML! And I was like "okay, maybe I didn't need to be worried?"
Then the actual LLM stuff happened and I was like "well... that's a dead dream." and now we can only hope we have the ability to turn all of those models off...
Upvote
54
(96
/
-42)
For all the good integrations I saw today - this one seems the most vexxing.
Their example of asking it for recipes was also an interesting choice. It seems to be one of the places ai consistently gets things wrong.
Their example of asking it for recipes was also an interesting choice. It seems to be one of the places ai consistently gets things wrong.
Upvote
78
(86
/
-8)
I also feel strongly about using AI in most of the way it's used. But I'm very stupid. My screen was acting up and I tried to fix it, so I used the best tool I have, my hammer. That didn't work out too well, obviously.
Maybe lots of folks use AI for everything like I used that hammer. But one in a while, there's a nail that does need to be hit in, and the hammer is actually good for that. There are some digital nails that need AI, too.
As long as it can easily be turned off, I'm happy to use it in some situations.
Upvote
4
(31
/
-27)
But how are they "inflicting" this on users? You can already run ChatGPT in a browsers, this just gives you the option to use it when you're making inquiries. As for "how many times will they ask?", my understanding is they're not "asking" anything, they're just putting a button where you can use it if you want. Not sure what you mean by "will the flag reset", unless there's a way to permanently turn off the option there's nothing to reset.
Upvote
69
(108
/
-39)
That’s my first thought as well. I want nothing to do with those data thieves.
Upvote
30
(63
/
-33)
Speak for yourself. I do want this. I don't want the OS to take screenshots and regularly send them to Apple but this sounds useful. Don't turn it on if you don't want it.
Upvote
37
(111
/
-74)
And one that can be answered perfectly well using a classic AI-free web search.
Upvote
38
(59
/
-21)
Upvote
66
(74
/
-8)
The screenshot attached to this story is of the OS asking if you want to use ChatGPT...
Upvote
52
(64
/
-12)
I was one of those who said “hopefully it’ll be a more thoughtful integration of ML than just cramming in a garbage-in-garbage-out chatbot like everyone else”. Ultimately they decided “why not both?”. At least it’s more akin to a search engine option you can just ignore, just like I currently ignore all the AI chatbot options I could engage with. But it still feels a little lame for Apple to cave and just throw in the same Wrongness Engine as everyone else.
Upvote
15
(29
/
-14)
ChatGPT is the scapegoat du jour and people here don't want anybody to have the choice to use it. Likewise they'll judge anybody who does or even those who have mixed feelings. You either hate OpenAI (but conspicuously Meta is never mentioned) or you STFU. That is the consensus here, astroturfed or otherwise.
Upvote
-8
(66
/
-74)
Adding onto this:
If there is no off switch, how many of these pop ups are we going to see? Will it just randomly pop up on screen when we’re trying to do some other task? Will it show up in a way that we might accidentally send data to OpenAI, like how we accidentally take screen shots of the Home Screen?
I already turn off basically all banner notifications outside of emergency ones, I don’t need ChatGPT being forced into my device against my will.
Upvote
9
(45
/
-36)
Everyone here seems to be panicking that this is some kind of unavoidable deep integration that hands your life over to ChatGPT.
I watched the actual presentation. They go on for, I don't know, 45 minutes showing off one thing after another that's dozens of small, task-specific machine learning things to help with/automate/enable one little thing after another - an awesome interactive calculator, searching your images by description, making automatic image collections, scoring images by how good they are, etc etc, it went on and on with small, specific ML tricks under the brand of "Apple Intelligence." And for all of that, they started off with saying how it's built privacy-first and is all on-device, except for some queries are served better on the cloud, so for that they built a new private cloud, where everything is sent and kept encrypted, it's verifiable by third parties, none of your data gets out.
Then at the very end they spent like a minute saying that if you ask Siri questions where it thinks ChatGPT could provide a better answer, it'll ask you, case by case, if you want it to ask ChatGPT. They'd already shown off a mountain of new Siri functionalities involving email, texts, calendar, reminders, alarms, photos, etc, that's all their own ML platform. It's just for weird things like asking Siri about recipes with certain ingredients, or how to decorate your deck, where it offers to go to ChatGPT, and you have to allow it.
So do you ask Siri for advice on getting your feng shui right and what to make for dinner? Because if you don't ask Siri these kinds of questions - and I'll note that if you do currently, you're a lunatic, because the current Siri would have no idea what you're talking about - then you'll probably never see one of these dialogs asking if Siri should enquire with ChatGPT.
Seriously, if you don't deliberately go about trying to use this new feature, you would probably never know it was there and never see an "Ask ChatGPT?" option. That's what everyone's freaking out about? A feature that's nearly invisible and would never have any effect on anything, or involve ChatGPT in any way, unless you go about asking Siri ChatGPT-type questions?
Many people are demanding a system-wide "turn it off" switch. There's no system-wide "turn it off" switch for, say, Google Search. Just don't use Google Search if you don't want to. I don't hear anyone screaming about that. Don't use this if you don't want it. It sends nothing to ChatGPT by default.
Upvote
210
(241
/
-31)
I'm personally not alarmed or enraged, but definitely skeptical of AI and disinterested in integration into devices that I use constantly throughout the day.
It's less about ChatGPT specifically and more that "AI" technology in general seems to be half baked. As long as it's wrong a statistically meaningful part of the time, it doesn't add value.
Add to that ethical concerns about how LLM training has been done - often without consent or respect to copyright. OpenAI specifically licenses Reddit content, the way for which was paved by Reddit screwing over big parts of their ecosystem and community. Then there's more ambiguous concerns, like what is done with voice recordings, queries I submit, or personal data needed to answer questions.
All of that for something I expect to add minimal value to some non-critical functions.
Upvote
21
(40
/
-19)
This puts me in a bit of a conundrum. On the one hand, I DO feel that AI tools can be very useful and increase the usefulness of many existing use case. On the other hand, I don’t know that I want OpenAI and Sam Altman to be the ones providin these tools to me. I wonder why Apple chose OpenAI over any of the less controversial competitors?
Upvote
-16
(13
/
-29)
1. I generally give Apple the benefit of the doubt, to the point I’ve had people insult me for not “properly” hating them.
2. You are absolutely 100% correct about this dark pattern and whoever approved it should be ashamed.
Upvote
18
(44
/
-26)
See Phat_tony's post before yours. Apple spent the vast majority of the "Apple Intelligence" section of the Keynote talking about specific, real world implementations of what they've always called "machine learning." None of them had anything to do with invasive chat bots or sketchy info or data being shared in an insecure way. They stressed that most of the processing happens on device, and when the software determines that the input is beyond the scope of same it is sent to Apple's purpose built servers with publicly verifiable security. This isn't just some random "let's put AI chat on everything", it's Apple leveraging their silicon/software integration to do some actually pretty cool new stuff. I'm hoping the Siri stuff works as well as they claimed, because that's an area where they are demonstrably behind the curve.
Last edited:
Upvote
71
(80
/
-9)
Upvote
4
(15
/
-11)
None of it changes my opinion. Maybe Apple will do it better and have features where it does add value, but the track record of AI is against them and they'll have to prove it before I give them credit for it.
Time will tell.
Upvote
-15
(22
/
-37)
The lengthy Keynote presentation that specifically addresses your concerns doesn't change your opinion? None of what you're worried about applies to Apple's implementation. They have given you the option to access ChatGPT, which you're free to ignore. I think you're conflating "AI" with "Chatbots", which if true I would share your concerns. But it looks to me like Apple has rolled out a bunch of really useful features that, but for the recent headline grabbing powers of things like ChatGPT, they would have most likely referred to as being enabled by "machine learning", their term of art for AI for quite a while now.
Upvote
32
(45
/
-13)
This! The commentariat here is the new generation of pearl clutching luddites.
This is the most harmless addition of a feature that's useful for some that I've seen and yet you wouldn't know it looking at all the pitchforks on display here.
I kind of feel sorry for them as it is as inevitable as day that AI will encroach on anything and everything in the future as it keeps improving.
Upvote
-3
(45
/
-48)
What is the money angle here? I don’t get it. How does apples deal with OpenAI work? Having ChatGPT on millions of Apple devices will be a massive cost for OpenAI. Right? So does that mean Apple is paying them?
But anyone can use ChatGPT for free now. So maybe OpenAI isn’t expecting to monetize the free version and are counting on Pro upgrades.
Apples deal with Google for default search was worth a fortune to Google because of the advertising revenue from each search.
For OpenAI there is no revenue for each inquiry to ChatGPT. For now. Has anyone heard of OpenAI plans to include ads in ChatGPT? I haven’t though it seems inevitable. With that in mind maybe OpenAI is paying Apple.
But anyone can use ChatGPT for free now. So maybe OpenAI isn’t expecting to monetize the free version and are counting on Pro upgrades.
Apples deal with Google for default search was worth a fortune to Google because of the advertising revenue from each search.
For OpenAI there is no revenue for each inquiry to ChatGPT. For now. Has anyone heard of OpenAI plans to include ads in ChatGPT? I haven’t though it seems inevitable. With that in mind maybe OpenAI is paying Apple.
Upvote
-2
(15
/
-17)