If Apple decides to outsource an AI project, I hope it’s OpenAI… 🤞🏻 Google sucks. 😝
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's iPhone AI Talks With Google and OpenAI Still Ongoing
- Thread starter MacRumors
- Start date
- Sort by reaction score
Crazy thought/ question just popped into my head… it was PORN that really drove the internet in its early days, same for bbs prior to that, and it was also PORN that drove home media (laser, beta, vhs, dvd, etc). So the question is this: where is porn in the larger discussion on ai? It’s gotta be there somewhere.
Not sure what AI porn can bring to the table? I mean there's plenty of real porn out there already, what exactly are you wanting AI to do? There must be 100 of millions of videos of real people hammering away at each other, I'm not sure how AI can make it any more realistic?
An on-device actually not-garbage AI, improved voice-to-text, these are things that would actually make me want to buy a brand new iPhone. Instead, I'll keep using my second-hand Xs for as long as it'll keep doing the same stuff that any modern phone can do.
Perhaps MR editors have realized that negative news drive way more clicks and comments (i.e. engagement) than positive news
If they colluded together to do charge more, that would be price fixing.
More likely is that they will compete with each other for access to Apple's business.
This is an extreme take.
It's like saying if you truly love nature, you'll dig a ditch in the forest and off yourself in it. That way, you will not consume any of nature's limited resources.
But that would obviously be an extreme take.
Fortunately, in the real world, there's often a middle road we can take.
Cloud computing and AI are significant and important technologies for humanity, and we can build them while being mindful of limiting their environmental impact.
The world isn't black and white, even if yours looks that way.
Certainly! I'm an NLP researcher and have a non cloud based LLM running on my iPhone ( actually three at the same time for OUTSIDE truth testing, vs INSIDE truth testing ).
I think it is starting to become clear that Apple’s approach to AI in the OS will be modular and layered.
There will not be a single ChatGPT-style AI model that does everything. Apple will continue to use AI for app features and expand on that. they will probably use a local AI like MM1 to be a front end to Siri that intercepts requests, interprets the intent and then finds apps and services that satisfy those intents. One of those services will probably be a chatbot, particularly for situations where Siri would currently respond with “here’s what I found on the web”. If Apple contracts with Google or OpenAI or someone else, then Siri would return an AI generated response to that kind of request. That kind of processing will certainly need to be one on servers. If the request is to generate an image, then the request would go to an AI like Gemini, OpenAI, or Midjourney.
For other tasks like “create a calendar entry for X on y date at z time” that will probably be handled by the local AI. For requests like “write a term paper based on ……”, That might go to a server-based AI which could be Apple’s or it could be a third-party. Which AI is used may change over time as the market shifts and as Apple decides which functions it wants to handle directly and which it wants to contract out to others.
this is like the early days of the web when browsers were new and only a few existed and larger companies had to scramble to build or buy one. Things changed rapidly and being early wasn’t a sure bet on winnning.
Don’t assume that this is just going to be a simple answer and a single solution.
There will not be a single ChatGPT-style AI model that does everything. Apple will continue to use AI for app features and expand on that. they will probably use a local AI like MM1 to be a front end to Siri that intercepts requests, interprets the intent and then finds apps and services that satisfy those intents. One of those services will probably be a chatbot, particularly for situations where Siri would currently respond with “here’s what I found on the web”. If Apple contracts with Google or OpenAI or someone else, then Siri would return an AI generated response to that kind of request. That kind of processing will certainly need to be one on servers. If the request is to generate an image, then the request would go to an AI like Gemini, OpenAI, or Midjourney.
For other tasks like “create a calendar entry for X on y date at z time” that will probably be handled by the local AI. For requests like “write a term paper based on ……”, That might go to a server-based AI which could be Apple’s or it could be a third-party. Which AI is used may change over time as the market shifts and as Apple decides which functions it wants to handle directly and which it wants to contract out to others.
this is like the early days of the web when browsers were new and only a few existed and larger companies had to scramble to build or buy one. Things changed rapidly and being early wasn’t a sure bet on winnning.
Don’t assume that this is just going to be a simple answer and a single solution.
Companies can only control their own footprint. They can choose the right partners but if they don't own the power hungry servers that provide services on their platform, they can neither take credit nor be blamed for the power used. It's just another form of offloading resource consumption and environmental damage to other countries or in this case, companies. Hopefully Apple aims to do more on-device AI than most competitors anyway.