Exactly. They don't care about privacy unless its a feature they sell you on.So, they are training on content that is common among users who don’t care about privacy.
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.
Here's How Apple is Working to Improve Apple Intelligence
- Thread starter MacRumors
- Start date
- Sort by reaction score
No way this is true. The rate at which these systems are improving is nothing short of incredible. To say something has peaked before it has even started is very shortsighted in my opinion.
Soooooo than what's taking one of the biggest tech companies so long just to have a usable assistant like Siri?
Let's all keep telling Siri, "Mind your own business, Siri, I'm sick of your half-breed interference, do you hear?", to see what happens.
The pattern over the last few months as it’s been pushed out is clear; from off by default, to on by default. From respecting my choice, to turning it on with each point update. From removing the data, to leaving 7GB of AI kruft on my device. It seems clear that the next step will be to have it on and you have to disable each part separately. The one after that to not let us disable it at all. Like Microsoft they are more and more insisting on jamming AI up our *** whether we want it or not.
Spell check on my iPhone drives me insane. Good example. This morning I typed "this" but a typo made it "thie". My phone just left it alone. However, I can type a real word and it will randomly change it to another word. WTF is that?
I wonder why only ChatGPT is an option with AppleAI.
It‘s like having only Google as option for search engines. It‘s strange for me.
Please add e.g. Grok and others, i don‘t like some ChatGPT results and hope i can try with others.
It‘s like having only Google as option for search engines. It‘s strange for me.
Please add e.g. Grok and others, i don‘t like some ChatGPT results and hope i can try with others.
I honestly find it all impressive, got me off guard from left field and can only see more and more insane advancements.
Just checking any random feed on ANY social media related to AI and honestly the results are just getting more mindblowing every single day… from full applications made on a whim to re-colored animated actors of yesteryears doing the performance of someone else to completely made up sound/music/images.
I don’t know what’s up with the “*scoff*, I’m unimpressed” general attitude towards it.
Or well, so be it, let’s let those that want to continue using their hands instead of the potential hammer of the future stomp away at nails the legacy way.
Can you expand what or why you don’t like ChatGPT specifically?
Nothing major for me, just curious as I find ChatGPT impressive as it is.
The latest models (and their mini variants) seem like they are kicking hard.
Indeed.
Sounds like way WAY more complex… plus the whole thing comes from “synthetic options” which then get filtered out choosing one that resembles the best one amongst them, repeat.
The less amount of iPhone users out there opt-in, the less speedy each iteration gets.
Sounds like a considerable longer loop than just “here, all of these are emails, thousands of terabytes of it”
I have yet to see one of those videos look even remotely good. Far from impressive, they’re derivative and flawed. Ripping off the work of others to shoddily re-skin a different body in an existing scene.
Maybe one day they’ll be good enough to produce convincing rip-offs, but there’s no creativity in a plagiarism engine.
I’m of the very personal opinion that everything always looks generic anyways. We being social beings is the natural state: to crowdsource generic acceptable responses.
Some of them are just good manners, like the way to end an email and signature (“best regards, x”; “thanks in advance, have a good day”; etc 1000x).
Heck the language we use is a crowdsourced agreement of sounds and definitions.
Others are like gravity or being pulled in a river, made stronger thanks to social media. Most will throw the same two or three sentences everybody else is writing, pertinent to a contemporary event, collect the likes, repeat afterwards with something else.
It is quite predictable the choice of language style and opinions, for example, we know ahead of time that the first responses in a MR forum post will likely be a short snarky take of not quite the main topic at hand (i.e it will divert to a political angle if any and associate the thing to oranges bad or a potato lost in space). Or the likes of “slow news day”.
All of this to say, for 99.9% of the emails you will need to write, an AI aided one will likely be fine and nobody will notice the difference; especially in not too distant future.
That odd email we write to our husband/wife/family, sure, I would proofread that at least and make it personal.
Writing to say that I do respect your opinion.
Also the usual reminder that this is the worst that it will ever be (likely, if the ensh!tifier dude bros don’t sink their teeth fully in it), since it has just started.
That said, this reminds of the “four chords of awesome” music: a lot of the famous catchy Grammy-winners top 10 billboards songs are indeed derivative, flawed, ripping the work of others and reskinning the tunes on top of new lyrics and mediocre arrangement patches.
So much so that these four chords of awesome musicians doing the presentation jump from one song sentence to the next sentence of a completely different song seamlessly. No AI involved in that.
My most disliked music style for that matter happens to be reggaeton (I’m Latin, just in case)… the background drums tunes is always exactly the same.
The skit of tons of songs using the same four chords, it’s great and hilarious:
Last edited:
I think it does. If Bob's AI is simple proxy to a larger AI back end then that's not a reasonable comparison to Apple's multi-tier structure, which is materially different. Just my opinion.
That has always been one of my favourite videos. It does show the underlying mechanics (for want of a better word) of western music, which reminds me that I often hear people complaining that music only uses four chords, saying it’s lacking creativity. We are used to a certain intonation and tuning, and certain chord structures just work better. That’s maths, physics, and cultural expectation, not a lack of creativity! You do have genres such as prog metal and jazz who push the envelope, but they’re not mainstream styles, and are just too much for a lot of people. I maintain that it displays a lot of creativity if you can make captivating music with fewer chords, rather than throwing the whole toolbox at your composition.
Anyway, sorry for the diversion, I’ll get off my hobby horse now…
I know generative AI will continue to improve, and I know that there will be legitimate uses for it, I just believe that after the hype bubble bursts, and people have gotten sick of it being perpetually inaccurate, it will settle down into being another piece of technology that makes the tools we use better. I strongly doubt that anyone’s ever going to just generate a movie by writing a series a complex prompts to describe each scene, but movies in the future absolutely will use AI tools in their CGI!
E.g. 1:
I asked ChatGPT to create a programmed file with xy parameters.
ChatGPT said it will create the file and when created i will be notified for the download.
After some time i asked when the download will be ready.
In 2 hours...
After 2 hours no file ready.
Again, ChatGPT said it will be ready in 2 hours.
Again not ready.
The next day i asked again.
Still said ready soon. Available in 30 minutes.
After 30 minutes still no download ready.
Again promised it will be ready in 10 minutes.
Then nothing.
I asked the next day again, but again ready soon....
I then asked many hours have past since i asked for the file the first time.
ChatGPT said it was only 4 hour ago.
So it is talking ******** or lying, what an AI never shoud.
E.g. 2:
Asked for code in Terminal.
Got the code, tried it but failures.
Asked again with hint to the failure message.
Then ChatGPT said, oh true. This is false.
Here is the right code.
This was false code again.
And so on.
So ChatGPT knows it answers with false code, and corrects it only when asked for it several times.
Complete useless.
Lost hours.
Enough examples?
I believe those are enough examples, but I also believe if you were hired to do a job that you should ACTUALLY know how to do it, versus asking chatGPT....and if your argument is that chatGPT is useless but Siri isn't, than you are completely insane lol