Advanced AI Dictation Not Enabled by Default in iOS 27 Beta

[maxfromdenmark]

Well it will come today in beta 2 !

Yes, Beta 2 is out now

 

[Johnstrass2]

This seems like a case of planned obsolescence on the Mac front. I have an M2 MacBook Pro Max with 96gb of ram - it seems quite odd that it's not powerful enough to run the same AI models as a base M3, unless Apple just wanted to limit the amount of machines they had to test it on.

I feel your frustration; I bought my M4 M=cBoo pro probably 6 months after your M2 device. OI bought my Apple Vision Pro in 2023, so that is DOA for AI..... I doubt this is "planned" obsolescence because few could have anticipated the type of computing requirements AI would need. That said, all of us should begin a new mindset to appreciate that the hardware lifetime-to-obsolescence is going to get ever shorter. MY AVP wont do AI, but the new refresh, being M5, will support AI.

 

[Johnstrass2]

I can’t imagine that the next generation of base iPad will have 12GB of RAM. This is another reason, combined with the much higher price of the iPad Pro, for the iPad Air models to continue to exist.

On-device or via a Cloud, it seems to me AI accuracy is going to become even more meaningful a differentiator. So a key question for the devices which don’t have this amount of memory is how good will the free or very cheap Cloud models they can tap into be? Apple hasn’t yet been forthcoming, for example, of what exact restrictions their Cloud tiers will impose. As OpenAI, Anthropic, Google, etc, all dramatically increase the price of their offerings to cover the actual CoGS, “ a good-enough default” offering by Apple at reasonable cost could take real market share.

I wish for two things form AI: 1) better diction with spelling and grammar aurtomaticlally improved. 2) integration of calendar, mail, Messages so that I can say "read my message from YYY about ZZZ and put that in my calendar". If both can be done "on device" and therefore with more privacy, then I will continue to be an apple FanBoy.

 

[mango316]

Hopefully we get this during the beta! I want this!!!

 

[Woodcrest64]

ChatGPT gets my voice perfectly, even when I stutter, on my “old” iPhone 14. It completely blows Siri out of the water. Idk why Apple didn’t do the same thing and instead is doing everything on-device hogging all of the local resources.

If the new voice recognition is only preferred 44% to the old one’s 17%, that means it must still be pretty bad.

I’d imagine if the same test was done with ChatGPT and current Siri, it would be 99.999% to .001%.

What gets me is Apple can upgrade its servers and everyone benefits where on device you are stuck with what you have unless you upgrade. Does that mean users can turn on and off on device dictation if the Cloud Compute ends up being superior overtime?

I have the same experiences as you with ChatGPT and Gemini. Its dictation is phenomenal compared to Siri. It’s not slow by any means for what I do.

 

[barkingsheltie]

This seems like a case of planned obsolescence on the Mac front. I have an M2 MacBook Pro Max with 96gb of ram - it seems quite odd that it's not powerful enough to run the same AI models as a base M3, unless Apple just wanted to limit the amount of machines they had to test it on.

My feelings too - Mac Studio M1 Ultra with 128 GB RAM - sometimes runs better than MBP M3 Max 128 GB! I saw a video a couple of years ago demonstrating Android Auto's ability to work with the microphone in a car vs Siri and CarPlay - the overriding message was Google was ahead in dictation in part due to training. This is true, it said, for Android on phones too. Google is Apple's partner here; I hoped some of these benefits would be shared across the board.

Evidently not, although this is one article, and the AA vs CarPlay may be a case of Apples vs Oranges, and there are deeper technical issues affecting dictation and its use on device

 

[CapitalIdea]

On by default in DB2 (at least on my unit)

 

[alchemistmuffin]

It’s also not available within macOS just yet, even for compatible machines.

It will be turned on at later updates, their focus right now is on mobile OS first, as dictation and Siri is likely going to be used on those more often than macOS.

Once it’s done for iOS and iPadOS, it’ll be smooth sailing for macOS users.

The new dictation model runs entirely on-device, so transcription quality stays the same whether or not the iPhone is connected to a network. It remains unclear whether the preview will stay off by default when iOS 27 is released officially later this year, or whether Apple will switch it on automatically at some point during the beta cycle this summer.

And this is first hint that Apple is trying to build an AI assistant that runs 100% on device. IN the end, they are planning to move this out of hybrid cloud, and move to fully on device thing, as this is found to be much secure. The neural engine is still not powerful enough however, and it will require more RAM to make this happen, but Apple is starting this move with M6 and A20 Chips which will have more RAM and much more efficient neural engine.

 

[Johnstrass2]

ChatGPT gets my voice perfectly, even when I stutter, on my “old” iPhone 14. It completely blows Siri out of the water. Idk why Apple didn’t do the same thing and instead is doing everything on-device hogging all of the local resources.

If the new voice recognition is only preferred 44% to the old one’s 17%, that means it must still be pretty bad.

I’d imagine if the same test was done with ChatGPT and current Siri, it would be 99.999% to .001%.

I would prefer to have as much AI done on device rather than having it go to a data center or who knows what is using it for a variety of purposes if Apple can at least have some of the ECAI done on device then I roaming support that goal for security reasons

 

[gnomeisland]

I just can’t believe an M1 Ultra can’t run it but an iPhone 17 Pro and Air can. I know, slight neural engine differences. I’m the last person to state such but if Apple wanted to support M1 and M2 systems they could. As macOS 27 is beginning to drop some support, I suspect Apple is preparing to sunset older M systems which hurts as they skipped the M4 Ultra and most likely will with the M5 otherwise I would have updated already. Ironically, AI will make the next iteration eye watering expensive compared to now.

I hate this timeline.

You may be right, even the M1 systems are enduringly capable. It could be a business decision to start sunsetting systems that are “too good”.

However, I wouldn’t call the difference between the neural engine’s slight (11 vs 35 TFLOPS)—most benchmarks put it at a real-world 60% speed increase. Not to mention the M3 was the first M-series chip that better integrated the GPU and NPU. That conceivably is the difference between real-time processing and too much latency given that the two differentiating feature are speech-to-text and text-to-speech which both work best in real-time.

 

[Lounge vibes 05]

Well it will come today in beta 2 !

Nope, it did not

 

[rjjacobson]

On by default in DB2 (at least on my unit)

Same on my IPad m4 beta 2

 

[Tagbert]

I opened the link and I don’t think people should be excited or jealous about this new dictation model.

In more than 50% of examples a human reviewer found the previous dictation model/capability as good or better than the new dictation model. There are about to be a LOT of underwhelmed people.

Going by the charts on that page... there are no tests where the reviewers preferred the old system. There were measures where "tied" was larger, meaning that the difference were not strong, but overall, this looks like a noticeable improvement.

Overall results

• Old model 17.6%
• Tied 37.7%
• New model 44.7%

 

[GlassFingers]

Pretty crazy that an M1 Max, 64GB will not be able to run this…..

 

[maccuser]

How do you enable it on the Mac?

 

[Macaholic868]

It’s in settings, under general, and then under keyboards.

Thank you!!!! Just turned it on in Beta 1 to mess with it while Beta 2 is downloading and installing.

 

[Macaholic868]

I feel your frustration; I bought my M4 M=cBoo pro probably 6 months after your M2 device. OI bought my Apple Vision Pro in 2023, so that is DOA for AI..... I doubt this is "planned" obsolescence because few could have anticipated the type of computing requirements AI would need. That said, all of us should begin a new mindset to appreciate that the hardware lifetime-to-obsolescence is going to get ever shorter. MY AVP wont do AI, but the new refresh, being M5, will support AI.

The M2 isn’t DOA for Apple Intelligence. You don’t get the most advanced on device models, like one that powers this feature, but the new Siri AI app still installs and is, IMO, just about as usable as it is on my iPhone 17 Pro that can use the most advanced on device models.

I’m sure that will change over time as more and more features are introduced that require those models, but it is what it is in that regard. The improvements you do get are still leaps and bounds better than things were previously.

I’m still getting used to the fact that Siri is useful now on my devices for things other than turning on and off our smart home lights. That is really all I used to use it for. That or making a phone call in CarPlay when I’m driving.

The experience on the M2 Vision Pro is no different than the experience on the M2 iPad Pro, the iPhone 17, iPhone 16, etc.

It’s still infinitely better than it was previously and it’s going to get even better as they continue to fix bugs and make refinements prior to the public non-beta release in September.

 

[zapmymac]

Turn it all off by default!!!

 

[maccuser]

I have turned it on, if it is on on-device processing, it is slow and just as bad as last siri transcription. Openai and gemini can transcribe can literally transcribe what you want to say before you say it, this is embarrasing for apple after all this time

 

[eifelbube]

Advanced dictation (AD) available on the Neo?
AD does not run on the iPhone 16 Pro because it needs 8GB of RAM. Neo and 16 both have A18 Pro chip and 8GB. That would lead me to believe the answer to my question is „no“. Am I overlooking something though?

 

[Dark_Omen]

I just hope Apple allows users to turn it all of as they can with 26. I’m sure future beta’s will add that ability… right? … right?!

I hope they don't just because there's people like you out there that refuse to want smartphones to be smart.

 

[johnsawyercjs]

How do you enable it on the Mac?

I think you can't, so far. On the iPhone, it's at Settings > General > Keyboard, then scroll to near the bottom (it'll be the next-to-last setting there), but on macOS, Keyboard has its own entry (not under General), and there are no settings there yet related to the new advanced dictation mode.

 

[clg82]

This seems like a case of planned obsolescence on the Mac front. I have an M2 MacBook Pro Max with 96gb of ram - it seems quite odd that it's not powerful enough to run the same AI models as a base M3, unless Apple just wanted to limit the amount of machines they had to test it on.

Of course this is the case, a greedy company wanting you to upgrade to the latest and greatest when in reality your older model would work just fine.

 

[GfPQqmcRKUvP]

Going by the charts on that page... there are no tests where the reviewers preferred the old system. There were measures where "tied" was larger, meaning that the difference were not strong, but overall, this looks like a noticeable improvement.

Overall results

• Old model 17.6%
• Tied 37.7%
• New model 44.7%

I don’t understand your point. Reviewers preferred the old model in 17.6% of instances and did not feel it was an improvement in 37.7% of cases. Together that means in more than 50% of cases reviewers felt the old model was as good as, or better than, the new model. The new model is undoubtedly better than the old model but given how bad the old model is I don’t think these results are encouraging.

 

[maccuser]

I don’t understand your point. Reviewers preferred the old model in 17.6% of instances and did not feel it was an improvement in 37.7% of cases. Together that means in more than 50% of cases reviewers felt the old model was as good as, or better than, the new model. The new model is undoubtedly better than the old model but given how bad the old model is I don’t think these results are encouraging.

Try it for yourself. It's processing dictation on the device; it's supposed to be noticeably faster than the cloud. But cloud dictation is so fast and accurate that Siri dictation is obsolete at this point. Very disappointed.