Apple Intelligence Features Coming to iOS 18 Beta Testers This Summer

[Atog]

The fact that the "Apple Intelligence" features are only available on the iPhone 15 Pro/Max is absolutely unnecessary, and pushing towards e-waste. Is has the same number of "neural cores" as the iPhone 14 Pro/Max ffs!

Yes, moreover this company is mentoring others for alleged eco-friendliness then they make almost all their iPhone lineup e-waste.

 

[anthogag]

Apple Intelligence is a good name for their AI implementation. 👏

 

[TechnoTiger3000]

Heard nothing about Apple Music AI. Sure hope they include it.

You can now ✨customize the color of the pay button✨!

 

[Macalicious2011]

i'm really disliking this whole thing going on right now in the tech industry, where they will announce something, then say "it's actually not done yet lol coming later enjoy the tease hehe"
or releasing half baked promises they won't deliver on until later.
hype you up to buy a product based on what they're promising and not what they're shipping.
it's time to stop!

This is becoming a pattern with OpenAI.

 

[topdrawer]

"apple intelligence"

apple = artificial?

 

[TechnoTiger3000]

Lol'd when they bragged about Siri not understanding ****.

 

[hagar]

What a huge disappointment. This announcement confirms Apple completely missed the boat on AI. Basically they announced what they want to build over the next years. Not what they will release in September.

And don’t get me started on this English Only limitation. It’s 2024. ChatGTP launched with dozens of languages from the start. International users will have to wait many years for features to trickle down.

Very very disappointing.

 

[rman0726]

View attachment 2387029
Think this is a good indication of how today’s announcements are perceived. Underwhelming to say the least. (This is CET so price drops started during the event)

I think it’s an indication, but not necessarily a good indication. For some reason people always seem to expect some monumental announcement that never comes. And the stock slightly dips. And then October/November comes around and it turns out they still sold a metric **** ton of iPhones.

People will be underwhelmed now and then Gen Z iPhone 15 pro/iPhone 16 owners will get their hands on new features and every other Gen Z iPhone user will be begging their parents for an upgrade because their life is over if they can’t create custom emojis.

 

[virginblue4]

The fact that the "Apple Intelligence" features are only available on the iPhone 15 Pro/Max is absolutely unnecessary, and pushing towards e-waste. Is has the same number of "neural cores" as the iPhone 14 Pro/Max ffs!

But that’s far from the full picture.

Here is the performance of the neural chips in each of the last 3 chips.

A15 - 15.8 TOPS
A16 - 17 TOPS
A17 Pro - 35 TOPS

The A17 Pro has over twice the neural engine performance compared to the A16!

 

[AbhijitD]

Well you do have:
Improved Safari
New Photos app
Reminders in Calendar
Better Control centre
Custom icons
Moveable icons
Custom lockscreen buttons
Better messaging/RCS support
Passwords app
Topographical maps
App lock on everything
Mail improvements
Game Mode (which may equate to JIT for emulators)
Machine learning notes improvements like recording transcription
Math notes calculator
Eye tracking accessibility

I wouldn’t call that useless.

Well apart from password app (which was much needed to counter the 1Password discontinuation of cloud based vaults) and Math notes caclulator (iPad only feature), rest of the stuff is meaningless to majority of the people. Btw, audio transcription is part of Apple Intelligence.

Secondly, the current generation iPhone and the current generation iPad mini can not use Apple Intelligence makes no sense to me. Hardware limitations are understandable but you can not even provide cloud support for current generation devices??

 

[kylelerner]

But that’s far from the full picture.

Here is the performance of the neural chips in each of the last 3 chips.

A15 - 15.8 TOPS
A16 - 17 TOPS
A17 Pro - 35 TOPS

The A17 Pro has over twice the neural engine performance compared to the A16!

This is extremely helpful, thanks!
My confusion was seeing, at one point in the distance past, that the A15 to A16 was minimal.

Still - would love to have the option to have the feature at slower performance.

 

[Kingcoherent]

The fact that the "Apple Intelligence" features are only available on the iPhone 15 Pro/Max is absolutely unnecessary, and pushing towards e-waste. Is has the same number of "neural cores" as the iPhone 14 Pro/Max ffs!

Same core count but lower TOPS. A17 pro is 35 whereas a16 is 17, so the newer chip is almost twice as powerful.

 

[Kingcoherent]

This is extremely helpful, thanks!
My confusion was seeing, at one point in the distance past, that the A15 to A16 was minimal.

Still - would love to have the options to have the feature at slower performance.

Ha. Beat me to it by minutes.

 

[dannys1]

View attachment 2387029
Think this is a good indication of how today’s announcements are perceived. Underwhelming to say the least. (This is CET so price drops started during the event)

The price drops of every single keynote regardless of what is annunced. Everything from the iPad calculator onwards was truly excellent.

 

[zorinlynx]

What a huge disappointment. This announcement confirms Apple completely missed the boat on AI. Basically they announced what they want to build over the next years. Not what they will release in September.

And don’t get me started on this English Only limitation. It’s 2024. ChatGTP launched with dozens of languages from the start. International users will have to wait many years for features to trickle down.

Very very disappointing.

I have to ask you, why do you actually want this stuff? So you can send your friends creepy looking AI images? I don't get it. Nothing about this "AI revolution" really speaks to me, other than better image and speech recognition, and stuff like the iPad calculator.

Getting computers to do our creative work for us is lame, not revolutionary.

 

[Jkeith85]

So nothing to really test big feature wise

 

[Mimiron]

Gating chatgpt on apple devices behind m1 chips and iphone 15+ series when it works in your browser. Apple must really think we’re stupid sheeple.

And i agree with elon. A lot of companies will ban os devices for security reasons due to native gpt integration.

 

[Atog]

This is extremely helpful, thanks!
My confusion was seeing, at one point in the distance past, that the A15 to A16 was minimal.

Still - would love to have the option to have the feature at slower performance.

The usable value of performance in TOPS for AI chips varies depending on specific applications and requirements. However, the following general benchmarks can be provided:

1. **Low performance (up to 1 TOPS)**: This performance level is sufficient for basic tasks such as simple speech processing, basic image recognition, or running AI functions in small devices like smartwatches or IoT devices.
2. **Medium performance (1-10 TOPS)**: This performance is suitable for moderately demanding applications such as real-time object recognition on mobile devices, advanced assistant functions (e.g., Google Assistant, Siri), and basic autonomous systems.
3. **High performance (10-100 TOPS)**: This performance is necessary for advanced applications such as autonomous vehicles, drones, augmented reality (AR), virtual reality (VR), advanced video analysis, and complex AI models running in real-time.
4. **Very high performance (100+ TOPS)**: This performance is required for the most demanding applications such as complex autonomous systems, large neural networks used in data centers, real-time rendering for advanced AR/VR, and other high-end AI tasks.


17 TOPS still plenty usable. Apple screwed us as they normaly do.

 

[Yujenisis]

View attachment 2387029
Think this is a good indication of how today’s announcements are perceived. Underwhelming to say the least. (This is CET so price drops started during the event)

No, this is a good indication of how a subset of the investor class (e.g day traders and some margin traders), view today’s announcement.

This a moment where I appreciate Apple’s conservatism. When the shallow glitz and glamor of unthinking “AI” LLM’s wear off (Google’s implementation has seen it receive the most criticism from the mainstream press in decades…), Apple will hopefully have a mature answer that can respect user privacy and actually offer accurate and personal solutions. But they squandered their head start.

The bit trying to make the handoff to OpenAI felt like such a decision mired in deep politics within Apple. Apple is ceding a tremendous amount to a third party, a massive compromise for Apple for Cook, but also perhaps a necessary one. This technology is so new and unpredictable they both felt they must include it but also the need to clearly bifurcate it from Apple in a way we have never seen. I expect there were meetings that could have made entire movies behind all of this.

I believe Apple utterly squandered the head start they were given in the acquisition of Siri in 2010. Siri was unique and predated the LLMs we’ve come to love and loathe, but was undeniably the first attempt at a true digital assistant. One wonders what vision saw the potential but failed to make the case that would have put Apple on the forefront here.

Exciting times. I am excited to see what Apple has in store, and don’t doubt for a second their “alliance” with OpenAI is simply one of necessity in the moment.

 

[Goldfinger]

But that’s far from the full picture.

Here is the performance of the neural chips in each of the last 3 chips.

A15 - 15.8 TOPS
A16 - 17 TOPS
A17 Pro - 35 TOPS

The A17 Pro has over twice the neural engine performance compared to the A16!

The M1 does 11 TOPS, so...

 

[one more]

“Apple Intelligence is free to use, but it is limited to devices with the A17 Pro chip or any M-series chip. That means to use these features, you need an iPhone 15 Pro or an ‌iPhone 15 Pro‌ Max, but the upcoming iPhone 16 lineup will support Apple Intelligence too.”

This was not very clear from the presentation, I thing Craig was talking about on-device processing at the time. Cloud-based AI should be doable on any device that can currently run ChatGPT standalone app, I guess?

 

[applemanit]

View attachment 2387029
Think this is a good indication of how today’s announcements are perceived. Underwhelming to say the least. (This is CET so price drops started during the event)

time to buy.

 

[Ctrlos]

Well apart from password app (which was much needed to counter the 1Password discontinuation of cloud based vaults) and Math notes caclulator (iPad only feature), rest of the stuff is meaningless to majority of the people. Btw, audio transcription is part of Apple Intelligence.

Secondly, the current generation iPhone and the current generation iPad mini can not use Apple Intelligence makes no sense to me. Hardware limitations are understandable but you can not even provide cloud support for current generation devices??

I reckon the majority of people just want a reliable device to text, surf and game. The customisation options will actually be the most important to the most people.

 

[MNTVJunkie]

Gating chatgpt on apple devices behind m1 chips and iphone 15+ series when it works in your browser. Apple must really think we’re stupid sheeple.

And i agree with elon. A lot of companies will ban os devices for security reasons due to native gpt integration.

So, the reason ANY browser can run ChatGPT is that ALL of the processing for that is done in the cloud. Obviously that has some pretty huge security and privacy implications, because it means every bit of text you send to GPT goes to a server somewhere (as does the text that comes back to you).

I'd imagine that Apple wants to do a lot of the AI processing on-device (something they have been doing more and more with Siri) so that Apple never SEE's your data. To do AI on-device requires the processor to do that work instead of sending it to a server to do it, but has the benefit of Apple NEVER having that data so it is not discoverable by government request, etc. Now, I could tell that not EVERYTHING they were doing required that, and it still doesn't explain why the M1 can do it but the A15 or A16 can't despite having MORE capable neural engines than the M1, I'm just clarifying the "well any browser can do it" statement.a

I've played with a local LLM myself, and running it on a 3080 (especially with larger LLM's) can still be pretty slow (although it isn't PAINFULLY slow, just "huh, that's taking a bit of time" slow). NVIDIA has pretty much come out and said that to be a good user experience you'll need something capable of 45 TOPS (for reference the NVIDIA has 15 TOPS and is a bit slow).

 

[TLuc]

Tells you how far behind Apple is in AI. Realistically, it'll be v1.0 in 2025.

Not available for immediate testing. Still in beta even when iOS 18 launches. No wonder they had to partner up with Sam Altman.

Yeah it's quite obvious this was put together recently. They are at least 1 year behind Google if not 2.