Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

MacRumors

macrumors bot
Original poster
Apr 12, 2001
67,826
38,458


Apple is in ongoing discussions with Google and OpenAI to use their generative large-language models as the backbone of new iPhone AI features, but no formal announcement is expected until the summer, according to Bloomberg.

iphone-open-ai.jpg

Today's report follows the news earlier this week that Apple has been in talks with Google to integrate its Gemini AI engine into the iPhone as part of iOS 18.

Apple CEO Tim Cook has promised that the company will "break new ground" in AI this year, and iOS 18 is expected to play a big role. However, Apple's internal work is said to be focused on smaller-scale AI features that operate on-device and do not require an internet connection.

To drive more powerful cloud-based generative AI features, such as the ability to create images and write essays based on single prompts, Apple is seeking a partner that has the necessary large-scale hardware infrastructure and compute capabilities already in place.

Siri is expected to take advantage of the artificial intelligence improvements that Apple plans to bring in iOS 18. One leaker has suggested that Apple is planning to use large-language models to entirely revamp Siri and turn it into the "ultimate virtual assistant."

Bloomberg's Mark Gurman says that one of the specific features Apple is developing is an improved interaction between Siri and the Messages app, which would let Siri auto-complete sentences more effectively and answer complex questions.

But with the likes of Nvidia and other AI companies continuing to reshape the technology landscape, some investors are reportedly concerned that Apple will fail to keep up with the pace of change, and may not see the "hypergrowth" expected of it unless the iPhone maker can boost its credibility by deploying generative AI features this year.

Article Link: Apple's iPhone AI Talks With Google and OpenAI Still Ongoing
 
  • Like
Reactions: superblockio
Can Apple like, avoid cloud-based AI? They can't say "Yay earth! Yay zero carbon footprint!" and partner with servers that demand so much energy and cooling water.
To be fair, a ChatGPT query only uses about the same energy as 15 Google queries. And BitCoin mining is estimated to use roughly 10,000,000 times as much energy per year globally than ChatGPT.
 
Last edited:
I thought Apple was developing their own AI. Looks like all they are going to do is partner with another company.

Why are they so behind here?
OpenAI and Google worked on their models for many years, employing the top researchers in the field, and applying tons of RLHF (reinforcement learning from human feedback — which takes time and large amounts of human feedback). Apple can’t just catch up that quickly for generative AI.

Look at how long Apple took to catch up on Google Maps with Apple Maps. And most people still think Google Maps is better in most countries.
 
I thought Apple was developing their own AI. Looks like all they are going to do is partner with another company.

Why are they so behind here?
The others are so far ahead. You can't release something so far behind the competition, so I am guessing Apple have no choice but to jump aboard someone else's train just to remain relevant in this space.
 
I thought Apple was developing their own AI. Looks like all they are going to do is partner with another company.

Why are they so behind here?

You think Apple can develop a better AI and Microsoft and Google?

In the meantime, Apple their competitors like Samsung already have partnerships with Google. So good luck to Apple if they want to compete against Android smartphones if they do it themselves.
 
Look at how long Apple took to catch up on Google Maps with Apple Maps. And most people still think Google Maps is better in most countries.
I agree. I would argue Apple Maps never caught up to Google Maps. Features like Look Around are still wildly incomplete in most of the United States outside of top 25 metro areas or major tourist destinations.
 
Apple is not known for moving into new areas quickly. They wait to see what the market is asking for but AI is moving so quickly that that approach is going to leave them behind. That said, many things are niche until Apple makes them mainstream so once this is added, it’s going to really change things. People like to say “AI” but most dont know what to do about it. They will later this year it seems.
 
Can Apple like, avoid cloud-based AI? They can't say "Yay earth! Yay zero carbon footprint!" and partner with servers that demand so much energy and cooling water.
They can but then will be very limited on their AI’s capabilities and then like how Siri is already one of the weakest voice assistants they would also have the weakest ai that people would complain about to no end. Im sure eventually as the chips gets faster the goal is certainly do everything on device but that is probably a long ways out. Like how Siri slowly became more on device. No matter what llm they use im sure they will make it the most restrictive however since people will try everything to make it say or do bad things in an attempt to make apple look bad or be able to sue them.
 
I also guess this partnership will be transitional, to avoid major hiccups on launch, as it was with Apple Maps and Apple Music, for example, where people could instantly compare it with competition.
 
  • Like
Reactions: Tagbert
I thought Apple was developing their own AI. Looks like all they are going to do is partner with another company.

Why are they so behind here?
I assume they are doing both. On device AI using the Neural Engine, but also off device LLM usage. If Apple is building their own server based LLM, that takes time. Just getting the graphics cards to do the training is hard right now.

Partnering with Google or OpenAI is the right move here.
 
The others are so far ahead. You can't release something so far behind the competition, so I am guessing Apple have no choice but to jump aboard someone else's train just to remain relevant in this space.
I thought Apple was developing their own AI. Looks like all they are going to do is partner with another company.

Why are they so behind here?
They are developing very capable multimodal smaller models that will probably answer 75% of all of your queries, and parter to outside vendors for the rest. Remember, apple business model is privacy based, so they will try to keep it local until they can, after this point, they will tokenize a index of part of your data if needed for the outside inference.
 
  • Like
Reactions: Tagbert
NVidia is owning the training hardware but there are new players on the inference side.

Apple has some levarage because of their NE, and unified memory but if I were they, I would open up the mac pro for outside nvidia GPU/NPU boards, just to be on the safe side.
 
  • Like
Reactions: nt5672
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.