I wonder what will come out when stupid (Siri) and smart ( ChatGPT) meet. It will be hard to believe Siri will function well with this integration. It’s going to take some time to make it work maybe iOS 19.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Preparing to Add ChatGPT Integration to Siri
- Thread starter MacRumors
- Start date
- Sort by reaction score
They'll probably need to otherwise the EU will come looking for more money for nothing.
Someone in Europe is doing well just now from EU hand outs.
Now if they add that to the HomePods too that would be great. Tired of mine being useless 83% of the time.
Wait a second… 83% of the time? So it’s NOT useless 17% of the time?!?! That’s amazing! How the heck did you pull that off? Techno black magic?
Using ChatGPT has become an important part of my job. It helps with everything from writing standard operating procedures to writing boring bash scripts to automate server admin tasks. Personally I used it to plan an inexpensive regional family vacation this summer and everyone had a lot of fun. It's really good at coming at problems from multiple angles that I wouldn't have thought of.
Lately I've been enjoying the o1 Preview model that seems a lot more comprehensive and seems to be especially good at programming, even if it is slower than 4-o. When a site goes down and there's something in the logs I haven't seen before, I'll feed it into o1 preview while I get to work on the usual tests/fixes. It's a great compliment to what I do, and also good at pair programming when I need it to review what I've done. It often finds ways to make things more performant and simpler. I've noticed steady improvements to multi-layered complex reasoning, and I find myself having to spend less time crafting specific prompts to get the result I'm after. I don't really notice it hallucinating anymore, and I'll occasionally test it with things I'm knowledgeable about. But I'm sure it happens, which is why I still use it with caution and review the output. I like that the o1 preview is pretty good at showing its work and explaining itself each step of the way. I'm really looking forward to the integration across all of my Apple devices! It's a great tool, when used responsibly.
Lately I've been enjoying the o1 Preview model that seems a lot more comprehensive and seems to be especially good at programming, even if it is slower than 4-o. When a site goes down and there's something in the logs I haven't seen before, I'll feed it into o1 preview while I get to work on the usual tests/fixes. It's a great compliment to what I do, and also good at pair programming when I need it to review what I've done. It often finds ways to make things more performant and simpler. I've noticed steady improvements to multi-layered complex reasoning, and I find myself having to spend less time crafting specific prompts to get the result I'm after. I don't really notice it hallucinating anymore, and I'll occasionally test it with things I'm knowledgeable about. But I'm sure it happens, which is why I still use it with caution and review the output. I like that the o1 preview is pretty good at showing its work and explaining itself each step of the way. I'm really looking forward to the integration across all of my Apple devices! It's a great tool, when used responsibly.
Ask ChatGPT: when is my baby sisters birthday?
The fact is that the domain of these machine learning systems is different, so framing one as more advanced than the other misses the point. Apple is clear that Apple Intelligence is about personal domain knowledge, and the other integrations will be for 'world knowledge' (for now, at least).
It’s marketing hype. Just like iPhones are “super secure!” It’s all marketing. And the data you give them they can see no matter what. I use ChatGPT and know how it works. Apple isn’t magically protecting anyone. And the whole point is one still needs to keep their ChatGPT subscription to do all they want. I am sure it will improve when Apple makes their own Apple Intelligence. But right now it’s ChatGPT integration and not Apple’s intelligence at all.
Except 48mp sensor in UW camera and an extra button, there’s nothing new in iPhone 16 Pro. So, why should Apple restrict software tweaks like photographic styles?
Apple says a lot about AI in iPhone and iPads in Swedish sites and we have no idea if we ever get it.
If Google and Samsung can offer AI in EU without any problems why Apple can’t? I can understand the demand from EU about opening up access to iMessage totally against Apple’s policy but regarding AI it’s the same for everyone. I can use ChatGpt in my Apple devices without any problems so, why Apple can’t allow us to use it as integrated functionality?
I can use Gemini in my Sony but can’t use Apple Intelligence even by changing region and language.
Apple says a lot about AI in iPhone and iPads in Swedish sites and we have no idea if we ever get it.
If Google and Samsung can offer AI in EU without any problems why Apple can’t? I can understand the demand from EU about opening up access to iMessage totally against Apple’s policy but regarding AI it’s the same for everyone. I can use ChatGpt in my Apple devices without any problems so, why Apple can’t allow us to use it as integrated functionality?
I can use Gemini in my Sony but can’t use Apple Intelligence even by changing region and language.
I hope I will be able to use GenMoji to create custom icons for the things I want to track with my AirTags in FindMy. 🤞
I swear to gawd this is a DAILY occurrence:
Me: Siri when's my next meeting?
HomePod: who's speaking?
Me: Radin
HomePod: Hello!
*me thinking about giving some scraps to my dog* yes/no
Me: Siri, is mango safe for dogs?
HomePod: I found some web results. I can show you if you ask again on your iPhone.
Tell me how to remove the feature systemwide. To believe, it’s being turned off because one “just” click a button would be rather naive thinking.
Last edited:
is visual intelligence pretty much the same thing as “visual look up“ , the feature that tells you what things are (that are in your photos?) If so, it’s pretty inaccurate. Hopefully it will be more intelligent than the feature that we already have, and I hope it eventually comes to iPad.
Lol. Maybe he should have quit with the slide rule. I'm old enough that I used one in college for physics class, and I still have it sitting in a box in my attic.
Gotta enjoy modern tech. 😀
Siri is utterly useless. It’s like having a conversation with a brick wall. I ask it for directions to a nearby Italian restaurant, and it responds with, “I don’t know where you are.”
And why the hell did I activate the location services for Siri?
No wonder why Apple needs help.
And why the hell did I activate the location services for Siri?
No wonder why Apple needs help.
Then you look back and discover it has made a fatal error. You tell it, it profusely apologises, then tells you a new answer that also has a fatal error. Ad inifintum. Don’t get me wrong I love AI but I would never trust any of the current free chat models with a maths based task. As the recent Apple research showed, even simple changes to the context of a question can give wildly varying answers, and that is exactly what I have found.
Are you saying we can use Siri without chatgpt? Or do we have to stop using Siri to avoid chatgpt?
man that inflation is a beast! + Siri has always been a useless assistant for me. Not that bright.
I fill out vendor apps by the dozens. I do not ever expect an AI system to butt in and say, "Here, let me fill that in for you."
That's because AI is simply too stupid to fill out a form. Hell, the things can barely do the OCR on a form. AI has no idea where to get the information from and where to place it on the form, what documents must accompany it, who must sign off on it, how to get it back to the right people, and even do ancillary things like requesting a certificate of insurance with the right coverages for the specific job being done.
AI has no hands so it can't even scan in a document if needed.
That's because AI is simply too stupid to fill out a form. Hell, the things can barely do the OCR on a form. AI has no idea where to get the information from and where to place it on the form, what documents must accompany it, who must sign off on it, how to get it back to the right people, and even do ancillary things like requesting a certificate of insurance with the right coverages for the specific job being done.
AI has no hands so it can't even scan in a document if needed.
At what point did you decide you didn't want tech companies monetising your thought process and why wasn't it 2003 with Google?
I work in the field, and while I have a scientific and financial interest in AI becoming successful, I do not share your optimism. Generative LLMs were fascinating around 2019 as they could generate text in a way that was inconceivable three years earlier. Current models are 10,000 times larger but not really that much better. They still suffer from the same fundamental flaws - they have no mechanism to grasp and work on meaning, and they have no understanding whatsoever of what they write. They "hallucinate" as people say, but that is just an inherent part of how they work, stringing together words (and letters) by statistic probabilities; there are no bugs to fix there.
Maybe they will eventually be good enough in many areas, I guess in some they already are, but I do not see them reaching "super intelligence", or AGI, or whatever the hype-department of OpenAI keeps promising to be just around the corner, for just a few more billions. They are unsustainable monstrosities, using enormous amounts of data and energy to solve minor problems with dubious reliability. Sam Altman is like some Victorian industrialist with a city-sized steam engine, promising us it will fly to the Moon if we just feed it enough coal. It won't, ever. The scientific basis for AGI is still decades away, and the required tech will have little in common with what we have now.
OpenAI and similar are running at a loss, users do not pay anywhere enough to keep them afloat, and the costs of training new models are rising exponentially. My hope is that LLM-tech will be good enough and sufficiently widespread by the time a big player like OpenAI crashes that the general populace will not dismiss AI as a whole, which could lead to another AI Winter.