Apple Intelligence to play catch-up to rivals across 2025

AppleInsiderAppleInsider
Posted:
in iOS edited October 20

While some employees believe Apple is currently lagging behind rivals its Apple Intelligence development, the company expects to gain ground in 2025 by introducing new chips and new models.

A person stands beside a large display showcasing various Apple software features, including writing tools, image playground, audio recording, and natural language search.
The rollout of Apple Intelligence promises a wealth of new and useful features. Image credit: Apple



Apple has been promoting the arrival of Apple Intelligence in its latest devices, which will roll out slowly across the next few months via software updates. Users will notice some new features and changes starting with Siri as early as iOS 18.1, which is expected to arrive by the end of October.

However, a Bloomberg newsletter claims that some employees feel Apple is significantly behind its rivals, who won't be sitting still on AI development either.

For example, internal reports at Apple are claimed to show that ChatGPT's responses to queries are currently about 25 percent more accurate than Siri. ChatGPT can also answer some 30 percent more questions.

Until that gap can be closed, Apple will emphasize features such as the promotion of timely notifications, summaries of longer emails and messages, and the partnership with ChatGPT for advanced queries. A more advanced Siri, generative emoji creation, and Image Playground are expected to arrive starting early in 2025.

Apple playing catch-up



Google and other competitors are already offering some features Apple will put into future Apple Intelligence releases, such as email summaries.

Apple's promotional focus for its AI thus far has been an emphasis on being able to do much of the work on-device, for greater privacy and security. For example, users are made aware when outside resources like ChatGPT are used, and what user information, if any, is disclosed.

Due to the high resource requirements of AI, the company has drawn something of a line in the sand. Only the iPhone 15 Pro and Pro Max or later models will run Apple Intelligence, and devices will need at least an A17 or M1 chipset, with at least 8GB of RAM.

Apple has a lot to do to close the gap between its own development and those of its rivals across 2025 and into 2026. This will be accomplished by hiring additional engineers, acquiring companies with promising new technology, and otherwise plowing lots more resources into Apple Intelligence development.

While Apple Intelligence matures, the company will focus on bringing out hardware that can run the advanced services as they arrive. The next iPhone SE is expected to have an A18 chip and 8GB of RAM in it, and the entry-level iPad is likely to get a similar update sometime in 2025.

According to the report, by early 2026, Apple's entire Mac, iPhone, and iPad lines will be Apple Intelligence-ready. The company will focus on using that, along with redesigns on some products and other updates, to convince users to upgrade.

Rumor Score: Likely

Read on AppleInsider

Comments

  • Reply 1 of 13
    XedXed Posts: 2,884member
    And, yet, these "rivals" will never be able to catch up to Apple's more ethical approach to how they deal with AI and personal information.
    9secondkox2chasmwilliamlondonssfe11appleinsideruserlibertyandfreesflagelbyronlwatto_cobra
  • Reply 2 of 13
    9secondkox29secondkox2 Posts: 3,139member
    I really liked what I saw when apple showcased Apple Intelligence. The only issue is that it’s not all here yet. 

    Apple has taken the concept and made it responsible, well rounded, capable, and polished. 

    The thing about ai up to now is that it’s kind of really janky. ALL of it. You never really know what you’re going to get. The quality just isn’t there. 

    Apple has invested the money into ethical “training” and has really gone the extra mile. It’s the only system that I’d actually trust and use with any expectation of reliability. 

    Apple has had a head start on everyone for a while with ML. 

    What some confuse for a lack of vision is actually just strong ethics. Where Apple sees a “do not pass” statue of Gandalf blocking the way into an unethical path, other just see that as a sign to step on the gas. 

    So now Apple has found a way to go that way, but take the high road. 


    ssfe11watto_cobra
  • Reply 3 of 13
    danoxdanox Posts: 3,414member
    Who are those rivals it can’t be Microsoft and Qualcomm they can’t even get their act together with the Surface Platform notice that Recall is still basically in dry dock, like all the other AI solutions. The Windows development kit that was supposed to help developers port Windows programs over to Arm has been canceled by Qualcomm, maybe Google is the one that is ahead?

    All of Googles mobile chips are at least five years behind Apple which makes it difficult to run AI on the edge, what’s Gemini doing these days (hallucinating) or maybe Meta has the solution to AI? All of Apple, so-called rivals have no real mobile/desktop hardware with any on the edge capability, all they have is super computers back home that will send you the answer on demand with a long tape delay.

    When you look at Apples rivals exactly what are they ahead in? I know it ain’t hardware, or OS mobile or desktop, and none possess multiple devices that can be linked together to work in any coherent fashion for the end user that says buy me at the upper end of the market.

    Apples, coherent presentation of Apple Intelligence basically provided a roadmap for their menagerie of confused rivals whose presentation’s changed after they saw Apples roadmap.

    https://www.digitaltrends.com/computing/qualcomm-mini-pc-dev-kit-cancelled/
    edited October 20 williamlondonthtwatto_cobra
  • Reply 4 of 13
    gweedogweedo Posts: 6member
    There is so much hype in this space right now. There is potential in the gen AI space clearly, but it is going to take some time to refine and focus the tech to be able to solve real problems. Billions upon billions being spent, uncertain if there will be a return on those investments. Exuberance is outpacing reality. These models still face significant challenges: accuracy and reliability, bias, and of course the costs and resource intensity. I expect Apple is working through all the noise to identify and build some useful capabilities.
    chasmwilliamlondonwatto_cobra
  • Reply 5 of 13
    chasmchasm Posts: 3,598member
    gweedo said:
    I expect Apple is working through all the noise to identify and build some useful capabilities.
    That has certainly been the company's focus, based on what they've presented us with so far.

    Sadly I lack any of the hardware to run Apple Intelligence just right now -- we'll see what the holidays have to say about that --  so for now I'm still stuck with what little HI (human intelligence) I still have left. :)
    edited October 20 williamlondonbyronlwatto_cobra
  • Reply 6 of 13
    12Strangers12Strangers Posts: 77member
    I’m seriously not trying to offend, but the previous comments seem to be from those who are not using AI on a daily basis or are a little behind on what the frontier models are capable of doing.
    williamlondonbyronlgrandact73
  • Reply 7 of 13
    chasmchasm Posts: 3,598member
    12Strangers said:
    I’m seriously not trying to offend, but the previous comments seem to be from those who are not using AI on a daily basis or are a little behind on what the frontier models are capable of doing.
    I think most of us are aware of what the bleeding edge has achieved, but apart from the novelty of chatbots and the bypassing-Google-via-CoPilot, I don't think the average computer user has really seen much out of AI that didn't seem like a novelty. That will change over time.
    thtwatto_cobra
  • Reply 8 of 13
    AppleZuluAppleZulu Posts: 2,181member
    It’s sort of funny how many patterns repeat right here in this space while people have no recognition of it happening again and again. 

    In this case, the repeated pattern is the lamentation and hand-wringing at how Apple is going to have to play “catch up” in a given space or product category. This angst continues even as Apple announces its initial foray into the area. Then, a couple of years later, somehow Apple has re-defined the thing through a careful, better thought-out implementation. Apple wasn’t first, but somehow within a few years, it’s everyone else who is trying to catch up to the altered paradigm. 

    So you know. Watch this space. 
    appleinsideruserwilliamlondonbyronlwatto_cobra
  • Reply 9 of 13
    kempathonnodgekempathonnodge Posts: 20member
    So far, Apple Intelligence is delivering well rounded features without passing anything to the cloud (without explicit consent). That is game changing, and I predict that other companies will be scrabbling to catch up with ‘on device’ AI for a limited number models so they can falsely convince users that ‘company x’ does on device processing, whilst the majority of their products will collect and sell the huge data mine that is people’s personal data. 

    Maybe these employees in apple are right to be frustrated, but I, for one, prefer a company to do what is right for the user, rather than progress without guardrails. 

    Keep going Apple, more haste less speed. 
    chasmthtbyronlwilliamlondonwatto_cobra
  • Reply 10 of 13
    wonkothesanewonkothesane Posts: 1,743member
    Except for in the EU…
    byronlwilliamlondon
  • Reply 11 of 13
    danvmdanvm Posts: 1,475member
    danox said:
    Who are those rivals it can’t be Microsoft and Qualcomm they can’t even get their act together with the Surface Platform notice that Recall is still basically in dry dock, like all the other AI solutions. The Windows development kit that was supposed to help developers port Windows programs over to Arm has been canceled by Qualcomm, maybe Google is the one that is ahead?
    Recall is included in the last Windows 11 update, which has additional AI features.  They already have Copilot in Windows, Edge, MS Office, MS 365, Teams, VS Code, Dynamics and some security services.  Additionally, AI services are available through Azure. It's clear that Microsoft is surpassing Apple in AI advancements. 

    And why you need a dev kit when there are devices in the market with Snapdragon Elite processors? I think Qualcomm did a great job with their first Oryon CPU for desktop.  They are ahead of Apple with 45 TOPS compared to Apple M4 38 TOPS.
    All of Googles mobile chips are at least five years behind Apple which makes it difficult to run AI on the edge, what’s Gemini doing these days (hallucinating) or maybe Meta has the solution to AI? All of Apple, so-called rivals have no real mobile/desktop hardware with any on the edge capability, all they have is super computers back home that will send you the answer on demand with a long tape delay.
    These super computers you mention is what Apple lacks today.  Se we could say Apple is behind Google and Meta too.  
    When you look at Apples rivals exactly what are they ahead in? I know it ain’t hardware, or OS mobile or desktop, and none possess multiple devices that can be linked together to work in any coherent fashion for the end user that says buy me at the upper end of the market.

    Apples, coherent presentation of Apple Intelligence basically provided a roadmap for their menagerie of confused rivals whose presentation’s changed after they saw Apples roadmap.
    MS is ahead of Apple with their enterprise / business ecosystem, desktop and cloud services. Google is ahead with cloud services and Android, who has a larger market share. Both companies have their AI services running and in consumers hands while we wait for Apple Intelligence.  Do you really think Apple is ahead of MS and Googe in AI?
    edited October 21 muthuk_vanalingamavon b7
  • Reply 12 of 13
    MarvinMarvin Posts: 15,486moderator
    danvm said:
    danox said:
    Who are those rivals it can’t be Microsoft and Qualcomm they can’t even get their act together with the Surface Platform notice that Recall is still basically in dry dock, like all the other AI solutions. The Windows development kit that was supposed to help developers port Windows programs over to Arm has been canceled by Qualcomm, maybe Google is the one that is ahead?
    I think Qualcomm did a great job with their first Oryon CPU for desktop.  They are ahead of Apple with 45 TOPS compared to Apple M4 38 TOPS.
    These are marketing numbers that don't hold up in real-world testing:

    https://browser.geekbench.com/ai/v1/84389
    https://browser.geekbench.com/ai/v1/84303
    https://browser.geekbench.com/ai/v1/77044

    M2 iPad is faster than the highest Snapdragon. Not to mention the results vary hugely depending on which product they are used in:

    https://browser.geekbench.com/ai/v1/52178 (single precision 401 vs 3428 on the M2 iPad)

    M4 is also Apple's lowest chip usually with no active cooling, they have M4 Pro, Max and Ultra, up to 8x this.
    13485williamlondonwatto_cobra
  • Reply 13 of 13
    danvmdanvm Posts: 1,475member
    Marvin said:
    danvm said:
    danox said:
    Who are those rivals it can’t be Microsoft and Qualcomm they can’t even get their act together with the Surface Platform notice that Recall is still basically in dry dock, like all the other AI solutions. The Windows development kit that was supposed to help developers port Windows programs over to Arm has been canceled by Qualcomm, maybe Google is the one that is ahead?
    I think Qualcomm did a great job with their first Oryon CPU for desktop.  They are ahead of Apple with 45 TOPS compared to Apple M4 38 TOPS.
    These are marketing numbers that don't hold up in real-world testing:

    https://browser.geekbench.com/ai/v1/84389
    https://browser.geekbench.com/ai/v1/84303
    https://browser.geekbench.com/ai/v1/77044

    M2 iPad is faster than the highest Snapdragon. Not to mention the results vary hugely depending on which product they are used in:

    https://browser.geekbench.com/ai/v1/52178 (single precision 401 vs 3428 on the M2 iPad)

    M4 is also Apple's lowest chip usually with no active cooling, they have M4 Pro, Max and Ultra, up to 8x this.
    It looks like there is a bug that results in inconsistent numbers in Qualcomm processors. Maybe that's the reason Snapdragon these numbers with Snapdragon processors.

    New Geekbench AI 1.0 Benchmark Analysis and Early Results - Signal65
    Geekbench AI and the State of the NPU - Creative Strategies

    Would be nice confirm if the tests are affected by the bug, and if it's the case, what are the numbers with the bug fixed. Even with the bug fixed, maybe Apple will be ahead. I'll keep an eye for the results in the Geekbench website.  
Sign In or Register to comment.