Is it ethical to use AI (other people’s work) and calling the final product your own work? If you do not want to upgrade, don’t. It is your choice. You will just have to wait to use some of Apples AI features.
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.
iPhone 16 Pro Max Apple is out of their minds. This is unethical to say the least.
- Thread starter NLLV
- Start date
- Sort by reaction score
Obvious click-bate.
OP never returned after their, probably paid-for, tantrum.
Artificial content for the sake of clicks is the REAL UNETHICAL behavior here.
OP never returned after their, probably paid-for, tantrum.
Artificial content for the sake of clicks is the REAL UNETHICAL behavior here.
I’m more concerned about battery consumption in relation to when “on board “ Ai is used. I got a feeling that bump in battery hardware capacity will still not be enough. We shall see.
This, I get the feeling that the efficiency gains on the new chip simply hedge against the inscreased power consumption.
I have an 15 pro that I chose as somewhat of a future-proofed device, might get used to carrying a power bank again.
Maybe when the image stuff hits, but honestly, the trivial text summary stuff that has been promised won't even register on your battery.
Apple is a really long way behind on this stuff. It's comedy that they couldn't ship it in iOS 18. Hopefully, they can catch up quickly.
Yeah, yeah, marketing wants to sell more phones, I get it.
Frankly, I don't fully trust Apple not to pull the same **** on my iPhone 16 Pro Maxes that are sitting in the cart. Fool me once.. etc.
I get it. My viewpoint is now that I am NOT on contract due to buying outright with the 14 pro, I get the best deal when trading in every year when I get the most for my current phone. AND continue to NOT be on ATT's contract.
Oh look another, "I don't understand this at all, so I'm gonna make a long post about how bad things are, despite not understanding any of it."
As soon as I saw the words "just fine" in OP's post, I stopped reading. Imagine if "just fine" were the metric of everything in technology and we just stopped when things were thought to be "just fine".
As soon as I saw the words "just fine" in OP's post, I stopped reading. Imagine if "just fine" were the metric of everything in technology and we just stopped when things were thought to be "just fine".
Indeed. Just shows how Apple is still primarily a hardware 1st vendor and software is always in catchup mode, don’t you think ?
Eventually, of course.
Apple has an on-device model with ~2B parameters using 4-bit quantization. The model uses about 2GB of memory. ~1GB for the model + ~1GB overhead. So you can't do much on-device. So there's an orchestration service that flicks the hard stuff to their PCC model in the cloud or OpenAI (in the case of some Siri requests).
The cloud model and even the maintenance of the local model over time need to be paid for. That's why it's ridiculous that Apple doesn't just come clean right now and say: "Hey this is a service guys", and make it available to everyone via subscription.
You are running a web browser (inside the app), running AI on servers across the country from you. Your phone can't run these models locally because there isn't enough RAM - RAM is the biggest limiting factor for all AI models, because you are just loading up a giant file of weights, and based on each piece of input it traverses a tree to find the next output.
Yes - your phone can run some "AI" locally - it already does. When you take a photo in portrait mode, a machine learning model is running to trace the outlines of your hair. That model has been optimized to run in the tiny amount of RAM available on your device.
Apple isn't just going to give you free chatGPT out of the goodness of their heart. Your third party AI apps will still work just fine when the 16 comes out. Going to chatGPT is the fallback for the local AI not being able to figure something out, they aren't going to make it the default. There is nothing "unethical" about not giving new features to a 3 year old phone.
I get that, but simply building sockets everywhere does not do for a "secure and private" consumer experience and would be a nightmare to pass to compliance in the EU.
Sure, but it's not "building sockets everywhere"; it's their own stateless zero-trust PCC which they are using anyway, with no OpenAI for the Euros.
Mystified by the original post. What is unethical about Apple offering something that you can voluntarily buy or not buy?
They've evidently decided that the new AI stuff is best supported on certain hardware. I assume they have some combination of technical and business reasons for this.
But they are in no way "forcing upgrades." If you have a better solution for you, stick with that. The market will work it out.
They've evidently decided that the new AI stuff is best supported on certain hardware. I assume they have some combination of technical and business reasons for this.
But they are in no way "forcing upgrades." If you have a better solution for you, stick with that. The market will work it out.
Maybe you could have actually watched the keynote and realized that Apple Intelligence can run locally. This requires more RAM of the phone.
I find this all a bit weird. Ars Technica on this:
arstechnica.com
"Most models that run on a local device still need hefty hardware," says Willison. "Phi-3-mini runs comfortably with less than 8GB of RAM, and can churn out tokens at a reasonable speed even on just a regular CPU. It's licensed MIT and should work well on a $55 Raspberry Pi—and the quality of results I've seen from it so far are comparable to models 4x larger."
Microsoft’s Phi-3 shows the surprising power of small, locally run AI language models
Microsoft’s 3.8B parameter Phi-3 may rival GPT-3.5, signaling a new era of “small language models.”…
arstechnica.com
"Most models that run on a local device still need hefty hardware," says Willison. "Phi-3-mini runs comfortably with less than 8GB of RAM, and can churn out tokens at a reasonable speed even on just a regular CPU. It's licensed MIT and should work well on a $55 Raspberry Pi—and the quality of results I've seen from it so far are comparable to models 4x larger."
This makes it even crazier that the new generation of iPhones has 8GB of RAM, when the model takes maybe 2GB and (looks at iPhone 14 Pro Max) 6GB is apparently not enough for Apple Intelligence.
This means that if Apple adjust the on-device model even slightly going forward, it's new phones all around, lads!
Maybe not strictly unethical, but cynical AF.
Yeah, it's Microsoft though. Apple "do it self".I find this all a bit weird. Ars Technica on this:
Microsoft’s Phi-3 shows the surprising power of small, locally run AI language models
Microsoft’s 3.8B parameter Phi-3 may rival GPT-3.5, signaling a new era of “small language models.”…
"Most models that run on a local device still need hefty hardware," says Willison. "Phi-3-mini runs comfortably with less than 8GB of RAM, and can churn out tokens at a reasonable speed even on just a regular CPU. It's licensed MIT and should work well on a $55 Raspberry Pi—and the quality of results I've seen from it so far are comparable to models 4x larger."
