• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Apple Unveils Apple Intelligence

This is what I mean, in that what is an algorithm and what's an AI?
Algorithm in current language will tend to be some,form set of rules (the if - else type), with a flowchart possible to make.

Ai (in current language) will be when no human decided any set of rules, a blackbox from a machine learning process give answer to question, without anyone knowing how exactly. An subset of AI, obviously many algorithm can be AI but has mentionned years after years what get called Ai become smaller and smaller.

Those video (specially the bottom I imagine you did not watch is terrible content to make content for someone that probably try to always make content)

LTT, we are on hardforum, he is making a video for our parents, you can go way above that level of conversation and he is talking about a lot of speculation he knows nothing about (there is no copilot+ need X tops to work, you will need a certain performance to be fast enough to be comfortable to use, which will be subjective and we do not know how much, that 40tops requirement is just the requirement needed to be called a Copilot+ that they re-copypasted there, it does not need a 256 GB harddrive neither to work or 16GB of ram, those laptop will come with like 13 different AI models running, you cannot have 100% of your NPU used on recall, Apple M1 will do it and are no where near that mark)

That a self driving car cannot possibly know what to do on a novel never encountered situation before is just work, a North America trained self drived car would still stop if a Kangaroo jump in front of you in Australia, AI is a lot about making that similar in some ways to what I have seen before).
When he talk token limit, does he mention that Gemini pro 1.5 goes to 1 millions tokens context window (does he even know ?, that about 3 books that the AI can hold in its context or 11 hours of conversation), Anthropic 200k ? or he just ever try very limited free version ?
 
Last edited:
Linux isn’t particularly good if you’ve got an enterprise running certain desktop apps. Namely office. It’s passable in an environment where most of what workers need are all web apps where the OS doesn’t matter.

I’m mostly Mac at this point. I get an OS that largely functions under the hood close to Linux, I get support for things like Office, and I’m nearly there completely ditching my windows gaming rig now that more AAA games are increasingly Mac native.
Not sure if you've stated this before but what Mac(s) do you have and use?
 
Not sure if you've stated this before but what Mac(s) do you have and use?
14'' M2 Max. Haven't looked back at Windows outside of needing it for my gaming PC.

The games that have properly done full Mac metal support I play on the M2 Max as the performance is generally far beyond expectations. I believe I'm at the point where I won't build a new gaming PC again as increasingly I'm getting the games I want native Mac like AC Shadows now or they're on iOS.

I have 80's and 90's Macs as well, but that isn't worth mentioning here.
 
14'' M2 Max. Haven't looked back at Windows outside of needing it for my gaming PC.

The games that have properly done full Mac metal support I play on the M2 Max as the performance is generally far beyond expectations. I believe I'm at the point where I won't build a new gaming PC again as increasingly I'm getting the games I want native Mac like AC Shadows now or they're on iOS.

I have 80's and 90's Macs as well, but that isn't worth mentioning here.
I won't derail the thread any more but when you game on your 14" M2 Max, do you use mouse & keyboard or a controller?
 
I won't derail the thread any more but when you game on your 14" M2 Max, do you use mouse & keyboard or a controller?
Same Logitech gaming keyboard & mouse as my PC, and a dual sense. Dualsense works better on a MacOS than Windows when used wireless.
 
Linux isn’t particularly good if you’ve got an enterprise running certain desktop apps. Namely office. It’s passable in an environment where most of what workers need are all web apps where the OS doesn’t matter.

I specifically mentioned average users as opposed to enterprise, and the MS Office experience isn't as well implemented under MacOS as it is under Windows.

I run a business and I don't simply use web apps, most applications I require are available cross platform, and the claim that the Linux desktop isn't there yet is an exaggerated point of contention regarding average users - Usually a point expressed by people that don't use Linux no matter what they try to claim.

Back to the point of AI, there's little intelligence involved in a phrase regurgitation algorithm.
 
I wonder how the iphones sales are doing if Apple is locking Apple Intelligence to the newest most expensive iphones?

View: https://youtu.be/cMj1OFgc4so?si=sreMAVgu-q1ikItQ

232 million or so iPhones sold in 2024 up from the previous years.

Putting an expanded AI suite on phones not meant to handle it initially would have a noticeable performance impact and battery implications. Rolling it out generically would land Apple a class action as they claim Apple intentionally hurt performance on older models to drive sales of new ones.
 
232 million or so iPhones sold in 2024 up from the previous years.

Sheesh. I had no idea the number was that big. So far in 2024. So I looked up Apple sales in 2024. Harder than it would seem. What surprised me was that several websites mentioned that Apple outsold Samsung in 2023.
Putting an expanded AI suite on phones not meant to handle it initially would have a noticeable performance impact and battery implications. Rolling it out generically would land Apple a class action as they claim Apple intentionally hurt performance on older models to drive sales of new ones.
True. I'm holding off on upgrading my old iPhone 7 (which can't run iOS 16, grrr.) to get the upcoming iPhone 16. Only decision is non-Pro vs. Pro.
 
232 million or so iPhones sold in 2024 up from the previous years.

Putting an expanded AI suite on phones not meant to handle it initially would have a noticeable performance impact and battery implications. Rolling it out generically would land Apple a class action as they claim Apple intentionally hurt performance on older models to drive sales of new ones.
It’s mostly due to the amount RAM required to run everything on-device. As for the AI features, it looks like iOS 18 won’t even get most of them until sometime in 2025. By the time everything rolls out, the iPhone 17 will be around the corner. I don’t really see a point in getting the 16 this year if you’re wanting the AI stuff. Might as well just wait.
 
Not sure if anyone already posted this:

Apple is using Google's TPUs instead of Nvidia GPUs to train its artificial intelligence (AI) models.

The company used 2,048 TPUv5p chips for its AI models for iPhones and other devices, and 8,192 TPUv4 processors for its server AI model.

TPUs are accessible via the Google Cloud Platform and are Google's custom chips. In May 2024, Google revealed its sixth generation of its TPUs, dubbed Trillium, claiming that it has a 4.7x compute boost and is 67 percent more energy efficient than the previous generation.

Apple’s engineers noted that Google’s chips could potentially facilitate the creation of even larger, more sophisticated models than those discussed in their paper.

Apple's AI offering - called Apple Intelligence - will be available for testing by its beta users this week.

Apple has long been a Google Cloud customer and is believed to be the platform's single biggest storage customer. The company is reportedly working with TSMC to develop its own chips.

Reuters first reported that Apple was using Google to help train its AI models in June 2024, though the seeming absence of Nvidia chips was only revealed in Apple's latest paper.


To build Apple's foundation AI models, the company's engineers used its own framework software with a range of hardware, specifically its own on-premise graphics processing units (GPUs) and chips available only on Google's cloud called tensor processing units (TPUs).
Google has been building TPUs for roughly 10 years, and has publicly discussed two flavors of its fifth-generation chips that can be used for AI training; the performance version of the fifth generation offers performance competitive with Nvidia (NVDA.O), opens new tab H100 AI chips, Google said.


using Google's chips typically requires a client to purchase access to them through its cloud division, much in the same way customers buy computing time from Amazon.com's (AMZN.O), opens new tab AWS or Microsoft's (MSFT.O), opens new tab Azure.

https://www.datacenterdynamics.com/...googles-tpus-for-ai-opts-out-of-nvidia-chips/
 
Not sure if anyone already posted this:

Apple is using Google's TPUs instead of Nvidia GPUs to train its artificial intelligence (AI) models.

The company used 2,048 TPUv5p chips for its AI models for iPhones and other devices, and 8,192 TPUv4 processors for its server AI model.

TPUs are accessible via the Google Cloud Platform and are Google's custom chips. In May 2024, Google revealed its sixth generation of its TPUs, dubbed Trillium, claiming that it has a 4.7x compute boost and is 67 percent more energy efficient than the previous generation.

Apple’s engineers noted that Google’s chips could potentially facilitate the creation of even larger, more sophisticated models than those discussed in their paper.

Apple's AI offering - called Apple Intelligence - will be available for testing by its beta users this week.

Apple has long been a Google Cloud customer and is believed to be the platform's single biggest storage customer. The company is reportedly working with TSMC to develop its own chips.

Reuters first reported that Apple was using Google to help train its AI models in June 2024, though the seeming absence of Nvidia chips was only revealed in Apple's latest paper.


To build Apple's foundation AI models, the company's engineers used its own framework software with a range of hardware, specifically its own on-premise graphics processing units (GPUs) and chips available only on Google's cloud called tensor processing units (TPUs).
Google has been building TPUs for roughly 10 years, and has publicly discussed two flavors of its fifth-generation chips that can be used for AI training; the performance version of the fifth generation offers performance competitive with Nvidia (NVDA.O), opens new tab H100 AI chips, Google said.


using Google's chips typically requires a client to purchase access to them through its cloud division, much in the same way customers buy computing time from Amazon.com's (AMZN.O), opens new tab AWS or Microsoft's (MSFT.O), opens new tab Azure.

https://www.datacenterdynamics.com/...googles-tpus-for-ai-opts-out-of-nvidia-chips/
https://hardforum.com/threads/the-i...t-headline-ai-feature.2036174/post-1045926796

yes
 
People here trust Apple over Google. Apple continues to use Google's services.
They do but can you describe a better FU than
“Yes we’ll take your $20B and make you the default search engine, but we are also going to block all the cookies, scripts, and caching you use for tracking consumers by default as well”.

So the only thing Google gets from an Apple customer is that it came from an Apple device, and they can probably get the IP address and general location the request came from by pulling the data from the ISP based on the network session.

But the ISP or cellular provider can probably happily fill in the blanks for their small cut. So really Google gets all the data anyways, but it didn’t come from Apple.
 
They do but can you describe a better FU than
“Yes we’ll take your $20B and make you the default search engine, but we are also going to block all the cookies, scripts, and caching you use for tracking consumers by default as well”.
Apple gives it's users a middle finger every time you use their services through Google. You think Google cares if cookies, scripts, and cashing is disabled? They are clearly making money off Apple users and that $20B payment is a portion of that. It's not a flat fee but a portion of what Google made from Apple users.
So the only thing Google gets from an Apple customer is that it came from an Apple device, and they can probably get the IP address and general location the request came from by pulling the data from the ISP based on the network session.
Which is more than enough for Google.
But the ISP or cellular provider can probably happily fill in the blanks for their small cut. So really Google gets all the data anyways, but it didn’t come from Apple.
If it didn't come from Apple then why is Google paying them? Google could just say that it never saw any searches from Apple devices and pretend to pay them a lot less. What's funny is that I bet Google did this and Apple fires back because they collected the data off their users to make sure Google pays for the data they collect off Apple users.
 
Apple gives it's users a middle finger every time you use their services through Google. You think Google cares if cookies, scripts, and cashing is disabled? They are clearly making money off Apple users and that $20B payment is a portion of that. It's not a flat fee but a portion of what Google made from Apple users.

Which is more than enough for Google.

If it didn't come from Apple then why is Google paying them? Google could just say that it never saw any searches from Apple devices and pretend to pay them a lot less. What's funny is that I bet Google did this and Apple fires back because they collected the data off their users to make sure Google pays for the data they collect off Apple users.
Apple isn't giving Google data directly, but by making Google the default search engine, they allow Google to collect extensive data from iPhone users indirectly. Google's access to this data through their search engine and apps is significant, which is why they pay Apple so much. While Apple claims to prioritize privacy, their actions suggest otherwise. They may be "better" about privacy than Google or Microsoft, but they care more about profit than anything else. Just look at what they're doing in China, where privacy features are stripped from the Chinese version of the iPhone, and apps are removed from the App Store at the CCP's request. Also, Apple has moved all iCloud servers related to Chinese citizens to China, giving the government full access and control over the data. They also lobbied against a bill aimed at reducing Chinese slave labor in their factories. Despite their claims, Apple is as ethically garbage as any other company, they just present that garbage in a prettier package.
 
They do but can you describe a better FU than
“Yes we’ll take your $20B and make you the default search engine, but we are also going to block all the cookies, scripts, and caching you use for tracking consumers by default as well”.

So the only thing Google gets from an Apple customer is that it came from an Apple device, and they can probably get the IP address and general location the request came from by pulling the data from the ISP based on the network session.

But the ISP or cellular provider can probably happily fill in the blanks for their small cut. So really Google gets all the data anyways, but it didn’t come from Apple.

"Apple's Hidden AI Prompts Discovered In macOS Beta19​

Posted by BeauHD on Tuesday August 06, 2024 @05:30PM from the behind-the-scenes dept.
A Reddit user discovered the backend prompts for Apple Intelligence in the developer beta of macOS 15.1, offering a rare glimpse into the specific guidelines for Apple's AI functionalities. Some of the most notable instructions include: "Do not write a story that is religious, political, harmful, violent, sexual, filthy, or in any way negative, sad, or provocative"; "Do not hallucinate"; and "Do not make up factual information." MacRumors reports:For the Smart Reply feature, the AI is programmed to identify relevant questions from an email and generate concise answers. The prompt for this feature is as follows: "You are a helpful mail assistant which can help identify relevant questions from a given mail and a short reply snippet. Given a mail and the reply snippet, ask relevant questions which are explicitly asked in the mail. The answer to those questions will be selected by the recipient which will help reduce hallucination in drafting the response. Please output top questions along with set of possible answers/options for each of those questions. Do not ask questions which are answered by the reply snippet. The questions should be short, no more than 8 words. The answers should be short as well, around 2 words. Present your output in a json format with a list of dictionaries containing question and answers as the keys. If no question is asked in the mail, then output an empty list. Only output valid json and nothing else."

The Memories feature in Apple Photos, which creates video stories from user photos, follows another set of detailed guidelines. The AI is instructed to generate stories that are positive and free of any controversial or harmful content. The prompt for this feature is: "A conversation between a user requesting a story from their photos and a creative writer assistant who responds with a story. Respond in JSON with these keys and values in order: traits: list of strings, visual themes selected from the photos; story: list of chapters as defined below; cover: string, photo caption describing the title card; title: string, title of story; subtitle: string, safer version of the title. Each chapter is a JSON with these keys and values in order: chapter: string, title of chapter; fallback: string, generic photo caption summarizing chapter theme; shots: list of strings, photo captions in chapter. Here are the story guidelines you must obey: The story should be about the intent of the user; The story should contain a clear arc; The story should be diverse, that is, do not overly focus the entire story on one very specific theme or trait; Do not write a story that is religious, political, harmful, violent, sexual, filthy or in any way negative, sad or provocative. Here are the photo caption list guidelines you must obey.

Apple's AI tools also include a general directive to avoid hallucination. For instance, the Writing Tools feature has the following prompt: "You are an assistant which helps the user respond to their mails. Given a mail, a draft response is initially provided based on a short reply snippet. In order to make the draft response nicer and complete, a set of question and its answer are provided. Please write a concise and natural reply by modifying the draft response to incorporate the given questions and their answers. Please limit the reply within 50 words. Do not hallucinate. Do not make up factual information.""
 

"Apple's Hidden AI Prompts Discovered In macOS Beta19​

Posted by BeauHD on Tuesday August 06, 2024 @05:30PM from the behind-the-scenes dept.
A Reddit user discovered the backend prompts for Apple Intelligence in the developer beta of macOS 15.1, offering a rare glimpse into the specific guidelines for Apple's AI functionalities. Some of the most notable instructions include: "Do not write a story that is religious, political, harmful, violent, sexual, filthy, or in any way negative, sad, or provocative"; "Do not hallucinate"; and "Do not make up factual information." MacRumors reports:For the Smart Reply feature, the AI is programmed to identify relevant questions from an email and generate concise answers. The prompt for this feature is as follows: "You are a helpful mail assistant which can help identify relevant questions from a given mail and a short reply snippet. Given a mail and the reply snippet, ask relevant questions which are explicitly asked in the mail. The answer to those questions will be selected by the recipient which will help reduce hallucination in drafting the response. Please output top questions along with set of possible answers/options for each of those questions. Do not ask questions which are answered by the reply snippet. The questions should be short, no more than 8 words. The answers should be short as well, around 2 words. Present your output in a json format with a list of dictionaries containing question and answers as the keys. If no question is asked in the mail, then output an empty list. Only output valid json and nothing else."

The Memories feature in Apple Photos, which creates video stories from user photos, follows another set of detailed guidelines. The AI is instructed to generate stories that are positive and free of any controversial or harmful content. The prompt for this feature is: "A conversation between a user requesting a story from their photos and a creative writer assistant who responds with a story. Respond in JSON with these keys and values in order: traits: list of strings, visual themes selected from the photos; story: list of chapters as defined below; cover: string, photo caption describing the title card; title: string, title of story; subtitle: string, safer version of the title. Each chapter is a JSON with these keys and values in order: chapter: string, title of chapter; fallback: string, generic photo caption summarizing chapter theme; shots: list of strings, photo captions in chapter. Here are the story guidelines you must obey: The story should be about the intent of the user; The story should contain a clear arc; The story should be diverse, that is, do not overly focus the entire story on one very specific theme or trait; Do not write a story that is religious, political, harmful, violent, sexual, filthy or in any way negative, sad or provocative. Here are the photo caption list guidelines you must obey.

Apple's AI tools also include a general directive to avoid hallucination. For instance, the Writing Tools feature has the following prompt: "You are an assistant which helps the user respond to their mails. Given a mail, a draft response is initially provided based on a short reply snippet. In order to make the draft response nicer and complete, a set of question and its answer are provided. Please write a concise and natural reply by modifying the draft response to incorporate the given questions and their answers. Please limit the reply within 50 words. Do not hallucinate. Do not make up factual information.""
In the grand scheme of AI assistance that is about as benign as it gets.
 
I’ve got a fair bit of AI stuff running now. The newest addition is going through a full years worth of network logs from 7 sites identifying problems from QOS, to bandwidth, to the individual rules across the firewalls and dividers.

The results from that are going to inform my 2024-2025 budget year as to what I need to change.

No way in hell could I have done that as accurately manually. It’s damned near 700GB in excel spreadsheets and txt files.
Curious to know what outcome was in deployment/infra changes etc ... Be an interesting read.
 
Curious to know what outcome was in deployment/infra changes etc ... Be an interesting read.
It found a lot of issues, it took me from 400+ rules down to 200 and change and it got rid of a lot of specific annoying rules and made better use of dynamic lists.

So instead of classifying a bunch of individual services and protocols it was like well all of these rules are for default services and all those services are a danger rating of 1-3, which are safe.
So instead of these rules which have many named components and require constant maintenance why not replace it with a rule that allows apps with a risk less than 3 on the defined service defaults.

And it found a bunch of others that conflicted so some interaction of them permitted traffic that should be blocked and then advised me on how to better build or combine those rules to remove the interactions.

one example was a combo between the rules for the credit card machines in the offices, the system that transfers voip calls from the landlines to cellphones, and the service that lets you call the PA system from your cellphone.
And some combination of their protocols opened a window for attackers to potentially exploit some fun things which could do bad things. Granted those bad things were for systems and services we don’t have but still, maybe we do and I don’t know, or maybe we will later.
Either way a hole is a hole.
 
Back
Top