Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Siri is not great, but there are many fields in AI where Apple is doing very good - fall detection, health, noise cancellation…
FYI, these aren't artificial intelligence. Implementations of Chat GPT or other LLMs are the most popular examples of actual "AI" that we've seen this decade so far.
 
  • Like
Reactions: locovaca
That's true ... AI will have entirely new and unexpected ways to be incorrect and unhelpful

An important distinction
:) I like this take, AI is doing things we didn't realize computers could do, which includes completely misunderstanding my request.

That said, Siri does actually get it right for me more often that it gets it wrong (not by much, but better than 50% of the time). When it does get it wrong, it usually provides laughs.

"I don't always screw up. But when I do, it is good for a laugh" - Siri
 
Did they? Or are we just taking rumors as fact now? And specifically what functionality is OpenAI going to be running?
They did, it has been reported as a fact, not a rumor.

They're surely going to use the GPT model like pretty much anybody else is doing now, and probably also DALL-E for image-related stuff, and all those conversions. I'd be surprised to see Sora or anything video-related specific to OpenAI.

According to rumors of what AI features we'll see in iOS 18, Apple needs that functionality :
  • Text-to-Text (GPT/GPT interactions)
  • Text-to-Speech (GPT/??? interactions)
  • Speech-to-Text (???/GPT interactions)
  • Image-to-Text (DALL-E/GPT interactions)
  • Text-to-Image (GPT/DALL-E interactions)
Although for speech related things, I'm not sure. Apple unveiled something pretty amazing in January 2023 which has been under the radar, and surprisingly hasn't even been implemented in Siri at WWDC 2023 : a better speech model than Siri, which gives the ability to read a book in their Books store. This AI model surpasses Siri by far.
 
  • Like
Reactions: sleeptodream
Apple can't even get Siri to work right. HomePods are a great example of pure failure. Apple expect us to think they can do AI better than the others? lol

Siri works great on my six HomePods to control 5 shades, 20 light switches, 2 plugs and and 5 TVs.
 
FYI, these aren't artificial intelligence. Implementations of Chat GPT or other LLMs are the most popular examples of actual "AI" that we've seen this decade so far.
Hmm... Artificial Intelligence turning Actual Intelligence would eventually become interesting. A bit of ethical potty training would be progress towards that.

Hmm... Potty training without a responsible parenthood...

I'll stick to Artificial for now...
 
Yeah, It's pretty good tracking congested tourist traps. It's pretty much something you'll get from a National Tourist Office. I think you pretty much made an argument for being cautious about this kind of tech.

They're not tourist traps. They are just great places which a lot of people visits. Would you say that places like Forum Romanum, the Colosseum, or the Vatican are tourist traps and you should never visit them if you go to Rome?

Yes, of course you can get this information from other sources or else chatGPT wouldn't be able to provide them to you. The great thing is that it comes to you in a condensed form without doing Google searches and reading several websites/web pages.

I'm going to Porto later this year. If I ask chatGPT "Provide a list of 5 Portuguese restaurant in Porto which isn't too expensive and link to their TripAdvisor review."

Here are five affordable Portuguese restaurants in Porto with their TripAdvisor links:

  1. Taberna Santo António: Known for its traditional Portuguese dishes and cozy atmosphere. TripAdvisor Review
  2. Nova Era Restaurante: Offers a variety of Portuguese dishes at reasonable prices. TripAdvisor Review
  3. Morfeu Marginal: Features delicious and affordable Portuguese cuisine. TripAdvisor Review
  4. Só Tapas: Provides an authentic tapas experience with a Portuguese twist. TripAdvisor Review
  5. Folias de Baco: A great spot for enjoying healthy and tasty Portuguese food. TripAdvisor Review
These options offer a good balance of quality and affordability, making them great choices for tourists.

It's so convenient and fast.
 
Yeah, It's pretty good tracking congested tourist traps. It's pretty much something you'll get from a National Tourist Office. I think you pretty much made an argument for being cautious about this kind of tech.

ChatGPT:
A tourist trap is a term used to describe a place or establishment that primarily exists to attract tourists and extract as much money from them as possible. These locations often offer overpriced goods, services, or experiences that are of lower quality than what might be found elsewhere. Tourist traps are typically characterized by the following features:

  1. High Prices: Goods and services are sold at significantly higher prices than their actual value.
  2. Low Quality: The quality of products or experiences is often subpar compared to what is available in non-tourist areas.
  3. Overcrowding: These spots are usually very crowded, as they are heavily promoted to tourists.
  4. Aggressive Marketing: There is often heavy marketing or advertising to draw in tourists, sometimes with exaggerated claims or misleading information.
  5. Generic Souvenirs: They frequently sell mass-produced, generic souvenirs that lack authenticity.
  6. Lack of Authenticity: Experiences and goods offered may lack the genuine cultural or historical significance that they claim to have.
  7. Convenient Location: Tourist traps are often conveniently located near major attractions, transportation hubs, or in highly trafficked areas.
While some tourists may enjoy the convenience or novelty of such places, others may feel disappointed by the inauthentic and commercialized nature of the experience.

Most of these places in Norway have high prices (like all places in Norway), high quality, a few of them might be overcrowded sometimes, authenticity and inconvenient locations.
 
1.2 is more than 1, therefore, 'decades' is the right word.

You say one and a half apples, NOT one and a half apple.

Nice try, but no cigar! 😁

If something has been going on for 12 years, you would never say (at least not correctly) that it had been going on for *decades*! That implies at least two, if not three, or more.

Even at first glance, the apples example doesn't work because there's a qualitative difference between a simple *count* of physical objects vs. a *duration* or *passage* of time.

Or, to use your apple example, if someone has owned a dozen and half Apple devices, we would never assert that he has owned *dozens* of Apple device as that would imply multiples of a dozen — as in, at least 24, 36, or more!

So, to distinguish this further, it's the difference between a measure that involves single things — one apple, two apples, etc. — vs. a measure that is a collective — that encompasses a particular quantity of things — dozens (12 in each group), decades (10 in each group).

Final example. If something happened 120 years ago, or 150, we wouldn’t that it occurred centuries ago. That would be a gross error, even a lie. We'd say it happened a century plus some years ago or a century and a half ago.

Bottom line. The deleted photos bug has not been there for decades. As MacIver noted, the iphone wasn't even around decades ago! 😁
 
Nice try, but no cigar! 😁

If something has been going on for 12 years, you would never say (at least not correctly) that it had been going on for *decades*! That implies at least two, if not three, or more.

Even at first glance, the apples example doesn't work because there's a qualitative difference between a simple *count* of physical objects vs. a *duration* or *passage* of time.

Or, to use your apple example, if someone has owned a dozen and half Apple devices, we would never assert that he has owned *dozens* of Apple device as that would imply multiples of a dozen — as in, at least 24, 36, or more!

So, to distinguish this further, it's the difference between a measure that involves single things — one apple, two apples, etc. — vs. a measure that is a collective — that encompasses a particular quantity of things — dozens (12 in each group), decades (10 in each group).

Final example. If something happened 120 years ago, or 150, we wouldn’t that it occurred centuries ago. That would be a gross error, even a lie. We'd say it happened a century plus some years ago or a century and a half ago.

Bottom line. The deleted photos bug has not been there for decades. As MacIver noted, the iphone wasn't even around decades ago! 😁

Yeah, I should've said "over a decade."
 
Yeah, I should've said "over a decade."
But even that assertion is wrong….! Let’s just move on, the bug is fixed, there isn’t a huge conspiracy that Apple is secretly holding our photos! It was a bug that affected backups and restoration of backups!
 
ChatGPT:


Most of these places in Norway have high prices (like all places in Norway), high quality, a few of them might be overcrowded sometimes, authenticity and inconvenient locations.
I'm saying that the output you get is a generic outtake of the standard cattle trail. It's like "list the 10 most recent singles by The Beatles". You don't acquire anything but surface knowledge which you would if you actually tracked them down yourself. Following your AI muck up will make a non geriatric tourist miss everything but the traps.

By the way, have the freakin' courtesy of writing your own arguments, you are not the first one being too impressed by "AI" and the easily recognisable output.

If you want to promote Norway to coastal travellers on wheels or wheels/ferries/++ starting out in Oslo with a good budget, stay at "Sommerro", have the "big menu" at Palace Grill, crossing the mountains make sure you visit Hogevarde and rent a trail bike. Skip the crowded fish market in Bergen, sleep to Molde. Or you can drive to Hoddevik and do some subarctic surfing.

Have lunch at Bud, drive the Atlantic road, then go Eikesdalen which is way more impressive than the "Trollroad". Time permits, visit Veidholmen and go kayaking/fishing and eat, then rejoin the coastal route (ferry) in Trondheim (Stay at Britannia and dine at Fagn or Credo before it closes and moves to Oslo) and stay onboard to Bronnoysund. Jump off and go to Vega for diving/kayaking/fishing/cycling/horseback. (One could also do the same in Rorvik for Vikna or Sandnessjoen for the islands).

Then go to Lofoten which will be terribly crowded if you must. Drive to Kvitnes farm for exceptional food, then end your trip in Tromsoe. Have a few beers at the most famous "bierstube" in Northern Norway. Mack has the best beer in the country micro breweries set aside. Wrong time of year, but Orca safari in the West fjord would be a treat combined with northern light winter time. If you have to tick off the northernmost point which is not Northern Cape but "Knivskjellodden", do so, but there isn't really that much to see north of Tromsoe unless one want to tick off that box.

As far as food goes, I've listed some that not only provides rather decent food (try to avoid salmon), but also opens up for a social experience. Travelling is always better when you get to know the locals. As far as the population goes, Bergen is a Hanseatic town which means it is very open to "strangers" like the small towns closest to the Atlantic Ocean south of Trondheim. Further north (Bronnoysund and onwards) it gets even better, in particular where there are few tourists.

Staying in Portugal, I'd rather visit Porto and the surrounding area towards the coastal border of Spain, or being a bit lazy, I'd go to Setubal. Fantastic seafood, and great beaches if the North Atlantic swell isn't too bad. Would also recommend the Duoro valley and have a taste of Quinta Do Noval for historical reasons. The survivor.

My apologies if my memory is a tad vague, it's been a while, but everything AI free. A way more authentic experience I'd argue.
 
  • Love
Reactions: Shirasaki
That's upto apple to invent that feature. They are selling user data (search queries) to google for 20 billion. They shouldn't be advertising privacy then, cause it straight up isn't true if they are accepting 20 billion from Google.
That ship has sailed a long time ago when Apple entered advertisement business and we saw App Store ads for the first time.
 
…how can something be both on device and query the web?
Twist: your device will always be connected, and all of your data is stored both locally and on the web. So, by doing “on-device” processing that sends query to the web, they are pulling the cloud copy of your data, not your local one, hence “your data stays with you”.
(Completely made up in case anyone wondering. Tried to be sarcastic but failed)
 
FYI, these aren't artificial intelligence. Implementations of Chat GPT or other LLMs are the most popular examples of actual "AI" that we've seen this decade so far.
FYI, these are exactly what AI is.

Fall Detection: an algorithm that detects a fall versus running down stairs is AI
Noise cancellation: an algorithm that detects ambient noises that should be cancelled out with other noise is AI
Health: algorithms that detect abnormalities in your personal trends is AI
Photo face detection: algorithms that detect and identify facial features is AI

If you want to get more specific, most of the examples above are machine learning, which is AI. Chat GPT is a subset of machine learning, which is a subset of AI.

Lots of people here don't even realize Apple just released 8 open model AI models that rival Chat GPT at multimodal activities like identifying not only what is in an image, but the context and being able to apply math to it.
 
  • Haha
Reactions: dk001
Twist: your device will always be connected, and all of your data is stored both locally and on the web. So, by doing “on-device” processing that sends query to the web, they are pulling the cloud copy of your data, not your local one, hence “your data stays with you”.
(Completely made up in case anyone wondering. Tried to be sarcastic but failed)
Decoding and understanding the query will be on-device. If the query specifically asks to search for real-time data, the query is made to the web. Either that or Apple relies on their already-'internalized' data sources for live activities (sports scores, flight info, etc). I say "internalized" because this data seems to already be either a part of Siri and Apple apps.
 
…how can something be both on device and query the web?

If the function you call uses device features it would be local. If the function you call needs to pull info from the web it queries online. Siri already does this but can be expanded more to provide a more natural chat.
 
FYI, these are exactly what AI is.

Fall Detection: an algorithm that detects a fall versus running down stairs is AI
Noise cancellation: an algorithm that detects ambient noises that should be cancelled out with other noise is AI
Health: algorithms that detect abnormalities in your personal trends is AI
Photo face detection: algorithms that detect and identify facial features is AI

If you want to get more specific, most of the examples above are machine learning, which is AI. Chat GPT is a subset of machine learning, which is a subset of AI.

Lots of people here don't even realize Apple just released 8 open model AI models that rival Chat GPT at multimodal activities like identifying not only what is in an image, but the context and being able to apply math to it.

Talk about redefining the definition of AI 😂

Fall detection is mechanical (electronic) with a built in timer that unless interupted triggers a predefined follow up. It is one reason I turn mine off as it will sporadically trigger when I am roughhousing with my dogs on the ground. Or I take my AWU off. That isn’t AI.

NC or ND in my Air Pods Pro is not AI.

Health is a “program” looking at data for a detectable predefined pattern that will trigger a response. Call it a glorified If/Then function if you want. While very detailed it is not AI.

Facial recognition/detection is not AI and has been around for a while. It has gotten better.

You need to realize an algorithm in not an AI. Simulating human intelligence (thinking like a human would) with problem solving capabilities is.

“Artificial intelligence is the science of making machines that can think like humans. It can do things that are considered "smart." AI technology can process large amounts of data in ways, unlike humans. The goal for AI is to be able to do things such as recognize patterns, make decisions, and judge like humans.”
 
  • Like
Reactions: locovaca
That's upto apple to invent that feature. They are selling user data (search queries) to google for 20 billion. They shouldn't be advertising privacy then, cause it straight up isn't true if they are accepting 20 billion from Google.
It’s not like you have to use Google, you can switch to any provider you want. That kind of cash just to be the default seems like something it would be hard for them to walk away from
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.