The recent Macrumors podcast had an excellent point: utilize the dang on-machine AI learning capability!!
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.
Report: Apple to 'Re-Examine' AI Development
- Thread starter MacRumors
- Start date
- Sort by reaction score
Lots of people conflating AI and Siri. Guess what - MS Cortana sucks too. Siri continues to be competitive with most the other personal assistant options mostly because they all suck at understanding and translating those into useful actions. Maybe its about to get better but most of them still fail. And Siri is about the same.
This is about more ChatGPT and Stable Diffusion type stuff. Which is fun...but it certainly has not produced "insight". Stable diffusion is going to put stock photos and some low level illustrators out of a job. ChatGPT is going to wow us and ....so far not much else. But the potential might be there.
This is about more ChatGPT and Stable Diffusion type stuff. Which is fun...but it certainly has not produced "insight". Stable diffusion is going to put stock photos and some low level illustrators out of a job. ChatGPT is going to wow us and ....so far not much else. But the potential might be there.
I have been saying that AI and voice recognition is the future and it seems like Apple just hasn’t cared. They have been too busy trying to milk every cent possible from their customers. I have a vision of how AI and voice recognition could absolutely revolutionize Apple and their products. The next version of macOS could look a lot like the advanced computers on Star Trek yet Siri has fallen behind. It should be so much more advanced right now then it is.
Typical DigiTimes article, nothing new, nothing for sure. Until Siri is made better don't even bother with AI.
Microsoft is not ahead or a leader in AI. Microsoft's version of ChatGPT has become a homicidal maniac! Don't forget Apple does not try to be first to market, all of the smart phones and tablets that existed before the iPhone and iPad are just a faint memory now!
It's only been available to a select group of the public for a few days, but Microsoft's new AI-powered Bing chatbot is making serious waves by making up horror stories, gaslighting users, passive-aggressively admitting defeat, and generally being extremely unstable in incredibly bizarre ways.
futurism.com
It's only been available to a select group of the public for a few days, but Microsoft's new AI-powered Bing chatbot is making serious waves by making up horror stories, gaslighting users, passive-aggressively admitting defeat, and generally being extremely unstable in incredibly bizarre ways.
Microsoft's Bing AI Is Leaking Maniac Alternate Personalities Named "Venom" and "Fury"
Stratechery's Ben Thompson found a way to have Microsoft's Bing AI chatbot come up with an alter ego that "was the opposite of her in every way."
futurism.com
Apple can't even get simple AI things right. Just take a look at the iPhone keyboard and how dumb and inaccurate the auto-correct is. I have no faith in them doing anything note-worthy in the AI wars to come.
Surely any intelligent technology humans create will understand itself to be in the service of humans and wholly exist to serve our purposes generously and benevolently. 😬Microsoft's version of ChatGPT has become a homicidal maniac! It's only been available to a select group of the public for a few days, but Microsoft's new AI-powered Bing chatbot is making serious waves by making up horror stories, gaslighting users, passive-aggressively admitting defeat, and generally being extremely unstable in incredibly bizarre ways.
Microsoft's Bing AI Is Leaking Maniac Alternate Personalities Named "Venom" and "Fury"
Stratechery's Ben Thompson found a way to have Microsoft's Bing AI chatbot come up with an alter ego that "was the opposite of her in every way."
Why Does This Horrifying Woman Keep Appearing in AI-Generated Images?
“Loab” is creepypasta for the AI generation, and nobody understands where she came from.
www.vice.com
Apple already does machine learning and neural engines.
The title “Will reexamine AI” is dim.
When or if Apple does LLMs they will do it their way.
They will do what they always do.
They wait and see how the competitor’s products crash, burn and end up in controversy and pain.
And then they will come out with their own version that is user friendly and a more focused product solely to make the devices work better.
The title “Will reexamine AI” is dim.
When or if Apple does LLMs they will do it their way.
They will do what they always do.
They wait and see how the competitor’s products crash, burn and end up in controversy and pain.
And then they will come out with their own version that is user friendly and a more focused product solely to make the devices work better.
“I can’t find my smarter on Apple Music.”
Not that I disagree with you on the general message. But AirPods were much more of a "big thing" than the Apple Watch or even the iPad ever were. And the new M* equipped laptops were also game changers in the industry.
He'll reserve Siri Pro for yet another tier of product.
And the existing tiers will get a new color.
I agree that the two aren't comparable, but I think people recognize that many of the underlying technologies and strengths of these AIs could be used to bolster the deficiencies in modern voice assistants. The two big complaints people have about Siri specifically are that it struggles to interpret complex requests and that it struggles to perform tasks that it wasn't explicitly programmed to do. ChatGPT by contrast does a decent job handling both complex requests and offering responses in ways it wasn't explicitly programmed for. (I could ask ChatGPT to tell me what would happen if the Borg from Star Trek invaded a planet of technologically advanced cats and the AI could provide a comrehendable response, whereas Siri would just default to searching the web). Merging this type of AI technology could make Siri more reliable and expand its range of skills without having to explicitly program new functionality in or install/manage a massive skills library like Alex.
(Disclosure, I did in fact ask ChatGPT what would happen if the Borg invaded a planet of technologically advanced cats and the results were fantastic. The cats won, if you were curious).
The root cause is the passing of Steve Jobs. Now they have a bead-counter as CEO and he follows the money.
The easy way to make tons of money is to make smal insremental improvments in a top selling product and and sell billions of copies of it. The iPhone is that. The "Big Announcement" today is a new color for the iPhone.
It used to be that Apple was ahead of the pack and others copied Apple. In fact MS Windows was born whrn Bill Gates saw a Macintosh and said to has people "If we can put THAT user interface on a PC, and sel it at a lower cost, peole will buy it." He was right.
Even Apple's OS. All of then MacOS, IOS and so on are based in BSD UNIX. I remeber when BSD UNIX was new. I first say it in the early 1980s and yes it was really astonishing how powerful and yet simple it was. It was a big step away from the other popular OSes of the time. What look at Apple, they stil use the same 1980's BSD under the hood. You think a trillion dollar company would have the means to advance that stat of the art in way you could not imagine. But no. They just make iPhone with more megapixels.
You could say the M2 processor is an inovation. Unitll you see what others are doing with ARM chips...
The link below has specs for an ARM-based processor chip with 128 ARM cores and 128 lane PCIe buss. Apple could use this to build a really nice Mac Pro. But it costs too much, their own M2 chips are cheaper and buying 3rd party chips eats into Aple's margins. Again the company is run by a Bean-Counter who looks only at the money.
But Apple absolutely does NOT own the high end of ARM processor ecosystem. They are some distance from the leaders. Look here: https://solutions-portal-cms-prod-b...a_Max_Rev_A1_PB_v1_05_20220728_8e27a0845f.pdf
If Siri is any reflection, they certainly haven't invested in AI.
Less snarky, ChatGPT has huge dangers associated with it. While search algorithms have bias, at least you understand the source material and can infer that bias by who is and isn't included. All that is obfuscated in the generative AI systems.
Plus, copyright is going to be a major problem for them. Is ingesting the entirety of the internet fair use?
Less snarky, ChatGPT has huge dangers associated with it. While search algorithms have bias, at least you understand the source material and can infer that bias by who is and isn't included. All that is obfuscated in the generative AI systems.
Plus, copyright is going to be a major problem for them. Is ingesting the entirety of the internet fair use?
Siri isn't the worst AI assistant (that dubious honor probably goes to Samsung's Bixby), but it's close. Which is a little extra embarrassing considering it was the first major AI assistant to be put on a phone. 😅
Honestly, some serious improvements in Siri have been long overdue; I'd settle for it being as good as the regular Google Assistant at this point.
Honestly, some serious improvements in Siri have been long overdue; I'd settle for it being as good as the regular Google Assistant at this point.
But GPT3 is not doing any thinking, when it tells you about the Borg invading a cat-world, it is just assembling words that have high statistical correlation to the input and training data. GPT3 does not "know" what a cat is.
In the example (Borgs and cats) GPT3 s very good because there can be no wrong answer. As long as the output is readable, it is good enough. But it does the SAME THING with real questions about science and engineering. Ask it to make scientific predictions based on some theory, and all it does is string words together that have good statistical matches to the training data. It produces readable but WRONG results more than half the time.
Search engines that produce readable but WRONG answers are worse than useless.
Apple simply bought Siri outright. As I recall the founders left soon after. And it hasn’t gotten substantially better since.
I guess Apple just needs to buy an AI company because it’s clear they don’t have the talent for it. Which is strange with how much they have promoted the “Neural Engine” but other than a little bit of photo editing they have never given a single example of what anyone has actually used it for.
Only about 10 years too late for that decision. I still don’t use Hey Siri for anything complex since all I get is meme skit material for conversation with my wife and friends.
Not really. They simply buy what they need. And its nascent times for how companies like Apple will use it. Exciting times.
Apple really needs to ring back the functionality that it keeps taking awy, like adding to notes by having Siri add it like it does to reminders ..by voice, and letting siri search photos again a as it used to be able to do. Siri could be such a help, but people don’t buy the iPhone for Siri, and so Apple has no impetus to make it what it could be. They just do what’s absolutely necessary to get by.
That’s what happens to companies once their primary focus becomes milking the customer for every dollar they can get. Innovation suffers severely and they fall way behind in the technology. Next thing you know the company becomes irrelevant and things just get worse from there.
Honestly, that's not really true anymore. So much research is public and so many of the crucial developments are open-sourced from day one, that it's relatively easy to get up-to-scratch quite quickly, with a good team. Apple can hire the best, so they could accelerate a strategy very quickly, I'd imagine. And actually, it was an Apple paper that led to the development of RWKV-LM—which we could be the basis for some significant movement in LLMs this year—so we know they're very capable of good work. I mean, I may turn out to be wrong, but there are good reasons to switch from the original Google/OpenAI approach to RWKV-LM (particularly if you're starting from scratch)... not the least of which is the potential offered by the way-cool SpikeGPT... 😎 Also, Apple has a sheeaaat-ton of $$$, so they'd feel no pain throwing a buh-zilleeon A100s at a problem... which, let's be honest, is one of the major hurdles... just simply raw compute power... which is just dollars... sad but true.
Apple has often benefitted from coming late to the party, so it could certainly happen again.