Apple and AI / LLMs [MERGED]

[Abazigal]

chatGPT isn't a Google product. Google has its own in-house AI solution.

If the "alternatives" are TOO good, and Apple doesn't have any, it makes Apple hardware a lot less appealing and not competitive. AI has the potential to make devices like the iPhone or Apple Watch obsolete if they can't take advantage of it.

AI isn’t confined to chat assistants. You have features like crash detection and ECG which have already proven to have saved lives.

At the heart of the issue is the way we interact with chatGPT (or its equivalents), which is pretty much a design problem.

Ultimately, I believe that people don’t want to have conversations with computers. We shouldn’t need to engage with computers via voice (or text) in order to pull useful information. Instead, digital assistants must become proactive to deliver information and data that we need before we ask for it. The more powerful the computer, the less talking, writing, and typing is needed.

Too many people in tech think that when we ask a digital voice assistant a question, we are yearning for a conversation as if we wish to talk to another human. This ends up being no different of a goal than wanting to spend time in the metaverse and make it come across as the real world.

Then there are the societal problems. For example, if ChatGPT were to lead to a decline in human-produced content, such a development wouldn’t be an accomplishment but rather a problem.

So rather than wonder what Apple could do to catch up, one should first ask themselves if chatGPT is even the right move moving forward. My attitude towards it for now is to continue to approach its potential with a great deal of scepticism, rather than call for Apple to also jump on the bandwagon.

 

[ipedro]

Option C - There’s a lot more to AI than chatGPT, and it may well go down the same way as Amazon’s smart speaker push. Makes for a cool tech demo, limited real world benefits, and ends up costing then more money than it’s worth.

Plus, knowing Google’s track record, there’s a non-zero chance they end up shuttering the service within two years.

Not forgetting that Apple makes the hardware that Siri runs on. It doesn’t matter how good alternatives are when they are not the default on devices like the iPhone or Apple Watch.

The more people claim that Apple is doomed if they don’t ape the competition, without considering if it even makes sense for Apple to compete directly, the more they tend to end up being dead wrong in the long run, I find.

There are technologies that define the next generation for a decade or more. The most recent was multi touch smartphones with an App Store. This transition effectively ended Microsoft's massive – and at the time unthinkable to shift – dominance of mainstream computers and went on to create a new competitive landscape of Apple vs Google. Google itself was left so flat footed that it took Android 5 years to finally catch up with an effective competitor, led by Samsung. It's easy to forget but Apple had the field virtually to itself from 2007 to at least 2012.

Nobody is saying that "Apple is doomed" – only that Tim Cook might be if he didn't anticipate this. If Apple misses the boat on this turning-point technology, then they'll be left behind trying to catch up and the iPhone will seem old news like Blackberries and ball cursor Android phones in 2007.

Conversational AI and UI are not a gimmick. It's not about having long conversations with a computer, it's about communicating like we do with one another – sometimes that's a quick question to a friend in the room and they answer you in natural language. Or you ask your colleague in plain English to find you some data that you both looked at last Friday, maybe it was Thursday, about that new sustainable bottle for your new product. They pick up on the context and remember it better than you did and know the exact name and details of what you kind of remembered.

This is what technology futurists have been predicting we were heading towards and would arrive and change everything. It's becoming clear that this is that moment. If you don't understand how natural human-like communication with computers that can parse complex information and context and answer virtually any question is the next big turning point, then wait and you will... eventually.

 

[Abazigal]

If you don't understand how natural human-like communication with computers that can answer virtually any question is the next big turning point, then wait and you will... eventually.

People don't want to talk to computers any more than they want to spend time in the metaverse. It's ultimately a means to an end - I just want the data you can give me; the conversation is simply the medium (and a fairly inefficient one too, I might add).

I will wait, and it won't be with bated breath.

 

[mxrider88]

chatGPT isn't a Google product. Google has its own in-house AI solution.

If the "alternatives" are TOO good, and Apple doesn't have any, it makes Apple hardware a lot less appealing and not competitive. AI has the potential to make devices like the iPhone or Apple Watch obsolete if they can't take advantage of it.

Exactly. After seeing Bing and Edge in action and having all sort of issues with macOS with iCloud not syncing properly and HomeKit driving me nuts in the past weeks, for the first time after a good 15 years of total devotion to Apple I thought if the time to move on has arrived. And coming from me is a big deal, I am a HUGE apple guy.


I want to be able to click on that buggy siri button on the top right corner in macOS and get a full generative AI that can compete with chat GPT and the new Edge/Bing capabilities at my service. Not at the end of the year with an ultra buggy rushed out release of macOS as the new apple standard dictates, I want it right NOW.

 

[Pakaku]

That's not really true, though. If you are having coffee with your friend, would you prefer using your voice to talk to them or would you rather sit at the table and send them text messages? If a waiter comes to take your order, would you rather just tell them what you want, or would you rather send them a text message while they stand there and wait?

You're looking at this through the lens of pre-AI technology, where the machine does a bad job of interpreting what you say/want. Imagine that the machine is indistinguishable from a human, and does just as good of a job.

People and machines are not the same thing... I'm fine talking to people, but I would never want my main interaction with a machine to be voice-driven.

 

[Shanghaichica]

Apple do not have an answer to these technologies. I think we just have to accept it for what it is.

 

[vinegarshots]

People and machines are not the same thing... I'm fine talking to people, but I would never want my main interaction with a machine to be voice-driven.

But what if people and machines were indistinguishable from each other?

 

[Abazigal]

Is anyone collecting bets yet on whether Google will shutter Bard within 2 years?

Google employees criticize CEO for “dumpster fire” response to ChatGPT

After Google's embarrassing event last week, employees take aim at CEO Sundar Pichai.

arstechnica.com

I don't think that's quite it.

The main 'problem' with something like this for Google is you can't run ads against it, so it's difficult to monetize in the manner that Google monetizes things.

OpenAI doesn't monetize things that way, so they are much more flexible with how they develop into a product.

Google should have developed it as a product if only as a hedge against something like this, while working on a business model.

I keep bringing this up: a market disruption is not a new technology - it's a new business model. That new business model is quite often now enabled by a new technology, but if you can't build a business model around a tech, it's not disruptive. Related to this is your go-to-market strategy - how do you take a product and position it in a market to get customers interested. They need to understand why they want to buy it, and then have it positioned in a way that customers will buy it. The iPhone's go-to-market was eased in a lot of ways because people were already carrying devices of comparable size that it could replace. Then it was delivering enough value-add over those devices to justify the price. Apple also experimented with the buying process but reversed that.

Google is frankly pretty terrible both at developing new business models and go-to-market. Generally they've just followed Apple, etc. And for services like Stadia, they completely ****ed up the go-to-market by having no reasonable solution for putting games on the platform, and launching anyway.

Google has great tech, but they never know what to do after that. The packaging, sales model, distribution channel, all that vanilla old-school business stuff - they suck. That's why the 'iPhone killer', etc claims are always laughable. You know how ****ing hard it is to be as good as Apple at that boring ****?

You know who's pretty good at it though? Microsoft.

 

[Topfry]

People and machines are not the same thing... I'm fine talking to people, but I would never want my main interaction with a machine to be voice-driven.

You mean like the homepod? Apple TV Searches? I dunno, if you flip the discussion, the one about user experiences, are people happy with the current Siri as a solution? I think not. Sad to say even Google’s assistant is considerably more reliable at this point. Natural and consistent complex voice interaction would certainly improve the user experience ( have ever got into a tangle of misunderstood commands with Siri?) ChatGPT sounds promising, and I wonder if it could be integrated with Siri (since Siri is basically a remote control and chat gpt a complex search engine)? Google used to be a third party maps and search engine on iOS, in the early days, perhaps Microsoft could do the same for Apple with Bing?
Few people want a computer as a “Her” style conversational buddy, but we would definitely appreciate an improved Siri interface.

 

[Shanghaichica]

chatGPT isn't a Google product. Google has its own in-house AI solution.

If the "alternatives" are TOO good, and Apple doesn't have any, it makes Apple hardware a lot less appealing and not competitive. AI has the potential to make devices like the iPhone or Apple Watch obsolete if they can't take advantage of it.

To be fair it’s not just Google Bard, whatever that turns out to be. Google have been ahead of Apple for years in this area. The google assistant is light years ahead of Siri. Google Duplex was unveiled years ago and it was only because people got their knickers in a twist that they had to Refine it. They also have really useful call screening features. Siri has improved but it’s not really moved on much from what it could do in 2011.

 

[rpmurray]

... We're talking about search engines where you ask one question and get the exact answer rather than scrolling through pages of search results ...

That's not where Google is skating. You may not have noticed but Google Search makes searches HARDER so you have to contend with ads and other sponsored fluff they can monetize. I expect their AI to be tuned to do the same thing.

 

[sracer]

chatGPT isn't a Google product. Google has its own in-house AI solution.

If the "alternatives" are TOO good, and Apple doesn't have any, it makes Apple hardware a lot less appealing and not competitive. AI has the potential to make devices like the iPhone or Apple Watch obsolete if they can't take advantage of it.

History has not proven that Apple having inferior alternatives diminishes the appeal of their hardware.

Apple has such a strong and devoted customer base that they don't have to have alternatives that are competitive (or even good). Siri has and continues to trail behind other "assistant" technologies (in spite of praise given it by the Apple faithful). Same for Apple Maps.

If Apple doesn't already have their own AI framework by now, then they're too late to the game. What will most likely happen is, just like they did with Siri in the assistance space, they'll buy some wobbly AI upstart and throw a ton of money playing catch-up until the next big thing.

 

[NT1440]

AI isn’t confined to chat assistants. You have features like crash detection and ECG which have already proven to have saved lives.

At the heart of the issue is the way we interact with chatGPT (or its equivalents), which is pretty much a design problem.

Ultimately, I believe that people don’t want to have conversations with computers. We shouldn’t need to engage with computers via voice (or text) in order to pull useful information. Instead, digital assistants must become proactive to deliver information and data that we need before we ask for it. The more powerful the computer, the less talking, writing, and typing is needed.

Too many people in tech think that when we ask a digital voice assistant a question, we are yearning for a conversation as if we wish to talk to another human. This ends up being no different of a goal than wanting to spend time in the metaverse and make it come across as the real world.

Then there are the societal problems. For example, if ChatGPT were to lead to a decline in human-produced content, such a development wouldn’t be an accomplishment but rather a problem.

So rather than wonder what Apple could do to catch up, one should first ask themselves if chatGPT is even the right move moving forward. My attitude towards it for now is to continue to approach its potential with a great deal of scepticism, rather than call for Apple to also jump on the bandwagon.

I’m with you, right now “AI” is having a giant marketing push. No doubt in an attempt to scare the **** out of white collar workers that their job is going to be replaced.

The reality, just like Machine Learning a couple years back was on the tip of every executive’s tongue, is that after the media coverage dies down a bit, is that these *tools* will find the right fit and become useful to augment work. But the promises will not live up to reality.

I had a call with some MS reps last week. I wanted to know whether a certain product MS introduced a year ago in preview was actually a product available for business now.

I **** you not, they pivoted my question into a demo of chatGPT in Bing. It gave AN answer of where in the admin portal to enable this setting. The reps were very pleased.

After the call I followed the transcript they shared….that feature doesn’t exist, let alone can be enabled. Even the documentation link provided led to a 404…

 

[chown33]

The current iteration of Bing Chat has reliability, credibility, or trustworthiness problems. It will authoritatively state things that are false, and that can be independently confirmed as false.

AI-powered Bing Chat loses its mind when fed Ars Technica article

"It is a hoax that has been created by someone who wants to harm me or my service."

arstechnica.com

In the several logged conversations discussing its original directives, as revealed by Kevin Liu, it repeatedly denies things that Microsoft itself has stated are true. Who're you gonna believe, the engaging chatbot, or those lying schemers at Microsoft?

These problems aren't confined to Bing Chat. Meta's Galactica chatbot seems to make stuff up on a whim:

Galactica: the AI knowledge base that makes stuff up

It seems like it's every couple of months that a big tech company releases a language model that's supposed to do amazing things. The Allen Institute for AI's Delphi [https://https://www.aiweirdness.com/stealing-a-giraffe-from-the-zoo-only-if-its-a-really-cool-giraffe/] , whose authors said it...

www.aiweirdness.com

In the words of that article's author, "Nothing Galactica generates is useful, because it's absolutely untrustworthy."

So while I would like to ask only one question and get an exact answer, I'd like the exactness to be real, not a hallucination.

Maybe we'll get there someday, but these are not the AI's I'm looking for.

EDIT
Here's a highly rated comment from the Ars Technica article:

Based on the aggressiveness of the response, unwillingness to admit fault, and debate tactic of claiming the source is biased, I bet a looooot of forum posts, tweets, troll content, and reddit threads are in that training data.

It's very rare that someone in a forum who is wrong says "oops, my bad, I was wrong". Either they disappear from the thread or get argumentative, which means most of the training data for "what to do after you've been shown to be wrong" from an online dataset will be never-back-down argumentative.

And a link from another comment:

Saturday Morning Breakfast Cereal - Artificial Incompetence

Saturday Morning Breakfast Cereal - Artificial Incompetence

www.smbc-comics.com

 

[Fraserpatty]

… Siri has improved but it’s not really moved on much from what it could do in 2011.

This is true, and where it has moved on, apple keeps backtracking. I am really upset that they took away Siri’s ability to search your photos for objects, dates, people etc. As someone with thousands of pictures, that was very helpful, much more helpful than searching for what I want in the spotlight or the photo search engine. I much preferred to use Siri. I also preferred it when Siri could add something to one of your notes. Now I have to scroll all the way to the bottom of a long list note in order to add something to it. Why does Apple keep taking away useful Siri functions? Makes it seem to get dumber instead of smarter. And I am one who likes to have Siri respond to me. With my HomePod and iPad when I tell Siri to turn on a light I don’t want just a tone played, I want Siri to say , such and such light is on like she did prior to iOS 16.2. When Siri respond verbally, it lets you know that Siri is doing some thing right. Otherwise, we are tending to forget that sometimes it does.

 

[droplink]

That's not where Google is skating. You may not have noticed but Google Search makes searches HARDER so you have to contend with ads and other sponsored fluff they can monetize. I expect their AI to be tuned to do the same thing.

I wish people would realise this.
Its about keeping people in the Google ecosystem.
Google search to lead to Google YouTube, to lead to Google Shopping etc. etc.

It look a while for people to realise that in Facebook, the users were the product. The same for Google.
Google is no more your friend than any of the other guys.

 

[norbinhouston]

Apple should just buy ChatGPT.

 

[ipedro]

Apple should just buy ChatGPT.

If openAI were for sale, Microsoft would have done so already, given that they've been continually investing billions into the company.

 

[Shanghaichica]

It will likely be reigned in, in light of all the scare mongering that’s going on

Bing’s A.I. Chat: ‘I Want to Be Alive. 😈’ (Published 2023)

In a two-hour conversation with our columnist, Microsoft’s new chatbot said it would like to be human, had a desire to be destructive and was in love with the person it was chatting with. Here’s the transcript.

www.nytimes.com

 

[RogerWilco]

Ouch.

Google shares drop $100 billion after its new AI chatbot makes a mistake

It wasn’t the “mistake” that caused the fall in share price, it was the realization that Google‘s management of this technology is a disorganized mess with no long-term strategy for implementing its use.

 

[GuillaumeB]

Just my two cents here but I don't really want to see ChatGPT baked into Siri.
Voice commands have not really taken off. i never see people around using it.
Do you really talk to your phone ?

I'd rather see Apple using AI deeply in macOS and iOS to boost Spotlight, generate security model based on the user activity or make API available to developers for their software (Midjourney-like APIs for image generator for Adobe-like software, APIs for music generators for Garaband-like software...)

 

[Abazigal]

it was the realization that Google‘s management of this technology is a disorganized mess with no long-term strategy for implementing its use.

So pretty much like every other google product of late? 😬

 

[ipedro]

If anyone was expecting an explicit announcement regarding Apple's response to chatGPT and Google's AI, they'd be disappointed as the keynote has come and gone. However, Apple did show us that they're in the game: they're now using a Transformer model in autocorrect and autocomplete across iOS, iPadOS and macOS.

This can also be seen in how Reminders understands how to sort groceries into categories that are organized logically for grocery shopping trips.

Finally, and kind of under the radar, Apple published a paper during WWDC in how it's building in the capability for developers to plug their own transformer models into Apple systems:

Deploying Transformers on the Apple Neural Engine

An increasing number of the machine learning (ML) models we build at Apple each year are either partly or fully adopting the [Transformer…

machinelearning.apple.com

Now, it's about ducking time for Apple to rebuild Siri to take advantage of this. It would tremendously impact the usefulness of HomePods, Apple Watch and AirPods. With Apple Vision Pro so dependent on Siri, you have to think that this is coming fairly soon – first half of 2024 if that's when Vision Pro is shipping.

 

[AustinIllini]

Apple is obviously working on something AI related but I also understand why they're careful to announce AI explicitly because AI in general has been met with immense skepticism.

 

[Fraserpatty]

Ok, but Siri is the consumer facing indication of AI, and Apple would do well to at least increase its abilities tenfold by the September event. It would be great if we didn’t even have to wait that long. Don’t call it AI, continue to call it, Machine Learning, whatever, but put Siri more centerstage.