5% divided by 12? 5% of what? Your math is as bad as Siri's
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.
More Advanced AI Siri Functionality Not Coming to iOS 18 Until 2025
- Thread starter MacRumors
- Start date
- Sort by reaction score
A more advanced Siri does not exist. Siri has been an unmitigated disaster.
That implies that you know it's wrong to start with. Which is the problem.
If I were the manager of an AI Team at the moment, I'd let my employees do whatever overtime they want to do.
And then, they should give the developers proper tools to be able to add various hooks throughout iOS to use Siri.
Apple has always given half-assed SDKs for Siri. It has never worked properly.
Overnight they could replace whatever language model they're using with GPT-4 and it would already improve Siri by magnitudes.
And then, they should give the developers proper tools to be able to add various hooks throughout iOS to use Siri.
Apple has always given half-assed SDKs for Siri. It has never worked properly.
I still just want PDF 2.0 support, complete CardDAV, CalDAV, and VCard 4.0 Safari to finally really stop website from asking to sent me notifications and more options for the default search engine. If iOS 18/macOS 15 would come with that I would call it a good upgrade.
It is a problem, but that's part of critical thinking. Just as we don't blindly trust whatever is online (or anywhere), we don't blindly trust LLM outputs. The issue is when people blindly trust and don't understand that LLMs can confabulate (make up) responses.
Siri can barely turn on my lights so this will certainly be welcome.
Not holding my breath but if Siri has indeed bern overhauled, I’m fine waiting.
As always, don’t rush it Apple, quality matters. But keep in mind you’re already running late.
I think we're thinking the same thing, Apple has millions upon millions of dollars in reserve. It's practically inconceivable of how many things they could be working on at one time. If they haven't been working on it it's because they didn't want to, not because they couldn't.
I'd add any mentions of his flipp'n hair.
This is SO FAR from anything Siri can do, I would just like it to not completely suck. It feels like incremental improvements have slowed or stopped for it for years.
Apple should just get rid of Siri entirely, until something far better comes out. It's a completely useless embarrassment.
You all do know that John G. -- the guy responsible for Siri since about 2018 -- is now leading the AI charge. Not sure what KPIs he's had the last 6 years, but they DO NOT include consumer satisfaction with Siri. Apple really, really needs to evaluate its top-level team. They are pretty much all the same set of guys, doing the same thing year after year. Nice club though re: $$$$ if you can get into it.
Seems like Apple is already starting damage control for those expecting "huge" update to IOS 18.....
Magic eraser, audio eraser, email summaries, dictation transcription and live translations to name a few.
But hey we have emulators now on iOS so whoopieeeee.
Yes… people SHOULD be aware of AI’s limits and take caution, not believing every answer at face value. But that’s not who we are as a society and as creatures, is it? The same warning could be applied to social media. But we have failed miserably at that, haven’t we? And the damage to our country and broader western society has been profound. AI will need stricter controls than “let the buyer beware.”
I wonder where they're gonna source all those gigawatts for the LLM cloud. Solar, Wind, Nuclear.... Coal? The LLM revolution is dramatically increasing the energy demands of large providers and shattering climate goals. Let's see if Apple really cares as much about the planet as they say.
Apple can push stuff off for years. The fact is that Siri is inferior to every other assistant out there. It can barely tell you the day and time. So they are going to incorporate AI, that is a good thing. However, in 2025 it will likely already be too little too late. I recently have ditched all my Apple HomeKit stuff in favor of Alexa. I didn't want to do that, but I am tired of waiting on Apple to get Siri where it needs to be. I have had so many problems using voice activation that it is a joke in our home. Yet the same request for Alexa is handled without issue. No Alexa is far from perfect, but a hell of a lot better than Siri.
Sonoma added one feature that I actually use: you can now make the list font in Finder bigger than 16px in accessibility settings. Yay!
Siri had a real moment, not just in the tech world, but in the cultural zeitgeist. They've really wasted it
The deal with openAI smells like panic. But I don't know if they have much of a choice. They have no hope of catching up. Even Google which has done much more in AI with deep mind was caught flat footed and making a hash of their search in reaction to openAI.
Nobody has solved the hallucination problem. Apple is going to have a real dilemma on their hand. On the one hand, they want to make a real leap with new look siri. On the other hand, they better be very conservative about how they use openAI because everything they do is heavily scrutinized and every misstep is magnified. Get ready for Siri-gate squared
Indeed which is where my problem starts. We introduce a problem where we need to consider the cost of interpretation and verification of results from an LLM versus the cost of doing the research and verification at the same time yourself. Large problems in verification are usually obvious if you have a lot of domain knowledge, but smaller problems are not necessarily obvious and might be nuanced. Not only that, how you ask the question can bias it and that is highly likely if you have little domain knowledge.
Your second point refers to my original one which is that the market promotes the utility of an LLM but carefully hides the negatives.
I like to think of an LLM as my good friend Paul. He's a guy who knows a little about everything but virtually nothing about any specific subjects. Yet his ego is larger than his critical thinking ability and is bolstered by people around him who know very little in comparison to him. In fact he actively surrounds himself with them. In the presence of those people he seems authoritative and trustworthy and has an answer for everything. But in the presence of experts in narrow specialities, which some of us are, the holes are very obvious to the point of him manufacturing credible sounding bullcrap. That, as you can imagine, somewhat destroys his credibility. Of course everyone is telling you all the time how clever he is and how much he knows.
And that's how the LLM market works. Relying on ignorance of others.