Apple Intelligence, intelligent path or fallen by the wayside?

If you listen to the markets, Apple is so far behind in the AI race that it will die in the next 5 years if it doesn't do something.

These statements are as old as time itself. Apple has been "about to die in the next 5 years" for the past 40 so all this blather by market pundits is nothing more than stock manipulation.

So is it true that Apple is so far behind in the AI race? Well, contrary to popular belief, Apple is not behind but has been a market leader for almost 20 years. Let me explain.

Back in 2009, Apple released iPhoto 09. Its main feature was called "Faces" and automatically created albums based on people's and pet's faces. At the time it was revolutionary. At first it had many errors, but the more that people trained the app, the more accurate it became.

Now, sit back and think about that long and hard will you. iPhoto's Faces feature took photos you had, scanned them, anazlysed them, and added them to an album based on that analysis. At first you had to do a fair amount of manual training, then it automatically did it with greater accuracy.

If that doesn't sound like what LLMs are doing today, then you're not paying attention to the technology behind AI. And remember, this was in 2009 Apple had that. Since then Apple has created data-detectors to automatically read text and recognise email addresses, phone numbers, calendar dates, location awareness, photo stitching, and all manner of things that are essentially AI in all but name.

Apple has also created numerous APIs so developers can take advantage of these abilities in their own apps.

So why does the market insist that Apple is massively behind in AI?

Stock market manipulation! They bet on rumours that they made up, and when Apple doesn't meet those unreasonable expectiations, they devalue Apple's stock.

For years Apple's efforts were called Machine Learning or ML. This worked into the market's hands because they had all decided that AI was the term they were going to back. Apple's continued use of the term Machine Learning instead of AI or Artificial Intelligence worked in the favour of the markets because it allowed them to preach that Apple is so far behind in the AI game.

So Apple rebranded ML to AI, but instead of calling it Artifical Intelligence, they called it Apple Intelligence.

Contrary to ChatGPT, Grok, Gemini, etc, Apple Intelligence has had almost 20 years of development going for it. But that matters not to the market because Apple Intelligence only came out in 2023. Despite the fact Apple Intelligence is built on top of the almost 20 years of ML development, it's a new product according to the markets and thus is well behind because it doesn't use LLMs.

So this market lie hides the immutable truth, Apple knows exactly what they are doing and when fully realised, they are going to kill the AI market and here's how they're going to do it.

If you look at the darling poster child that is NVIDIA, AI is going to require huge amounts of power and resources to make AI work. If you do look at that then you are an idiot. Mark my words, the hullabaloo over NVIDIA we currently see as our processing requirements for AI increase will be over in less than 5 years. In fact, the markets are already starting to preach the AI bubble is about to pop. But you know who will still be around when it does pop (and mark my words, it WILL pop)? Apple.

So why will Apple win out and NVIDIA will ultimately fail? It has to do with the approach that Apple has made since 2017. Neural Engine.

Since the A11 in the iPhone 8 and 8 Plus, Apple has had a dedicated chip for Machine Learning or AI purposes. Each year this chip gets bigger and more powerful in order to meet demands required by AI. But why does it exist at all? Because Apple has chosen to make all their devices capable of doing AI stuff. Apple has chosen to do it all on device instead of on servers. What this allows Apple is a smaller datacentre, less draw on the power grid, and far higher security than ANY of the AI developments of their competitors. It means that if a device needs more heavy lifting, THEN and only THEN will it upload to a server to process that data.

While Apple does allow 3rd party tools like ChatGPT to plug into Apple Intelligence, it still does most of the heavy lifting on the device. That means iPhones, Macs, iPads, Apple Watch, Apple TV, Apple Vision, etc are all doing the heavy lifting and the servers are just giving a little boost.

By having a dedicated chip, or more precisely, part of a chip, Apple’s AI doesn’t need as much power as its competitors.

Given that SOCs are starting to become a thing even in the Windows PC world, the more SOCs that have a neural engine type of functionality, the less power will be needed on the grid.

This leaves NVIDIA in a position of death. Currently many AI efforts need huge amounts of processing power and they get this currently through the use of graphics cards as the GPU has more computing power than a CPU. But if this is being done on device, then NVIDIA hasn’t got a showing because they don’t make SOCs. This is why I say that in 5 years NVIDIA is not going to be in the position it is now. Of course, NVIDIA isn't going anywhere. It's still needed for gaming and crypto, but even that is becoming less of a given these days. Remember, the first outing of the M1 Ultra made by Apple, trounced a $50,000 NVIDIA graphics card in almost all areas that matter. Next year we should see the M5 Ultra if the release schedule is maintained.

Doing AI on device adds a level of security that can't be achieved on a server. Remember when Siri started on devices and it uploaded the commands to Apple's servers? There was a lot of hoohaa over the fact that Apple could listen in on your requests, not that there was any evidence that they did for any reason other than training Siri. When that all moved ondevice, it meant that no one could listen to anything you did.

The same thing will happen with AI, specifically Apple Intelligence. Only really heavy lifting will be moved to Apple's servers, or to ChatGPT via the plugin. With no uploads to the server, there is no chance of 3rd parties reading and learning from your searches. Add to the fact that Apple does not sell your data (despite the claims), doing as much as possible on device is the only logical move.

Apple doesn't play by anyone's playbook other than Apple's. They might look at what others are doing and make their own foray into that market (iPhone, iPad, Apple Watch, Apple Vision, Apple TV), but mostly they'll release a product when its ready and on their own terms, not when markets dictate when they should. It's worked for Apple for over 40 years, why change a good thing?

Where Apple could be said to be behind is in the type of AI that isn't overly great in the first place. Things like image creation. Image Playground is a gimmick, but I don't think it's going to stay that way. Especially now that Apple owns Pixelmator. I expect GarageBand, Logic Pro, iMovie, and Final Cut to get some form of AI abilities when they're ready. Siri might get more of a ChatGPT type interface but I'm not sold on that being the way Apple should go. A voice assistant isn't actually as useful as people make it sound. But then those people are gaming on the idea that we're going to be talking to our computers all day every day. Hint: It won't happen.

I would like to see Siri have more of a Copilot interface, so the search box takes over. I feel that's more useful. Of course Microsoft has killed any real utility with the Search bar in Windows and then decided to add a separate floating window for Copilot which is an odd choice. But then it is Microsoft.

So all in all, Apple is so far from being behind in AI. It's chosen a more reliable path than any other AI company. Sure, this doesn't fit the market idea of AI but the market doesn't understand the engineering side of AI so are a worthless sounding box any way.

Apple's approach is to give users AI without users knowing they're using AI. Which is 100% the way an operating system should be. Linux gives users more control than they want, and Windows gives users too much control in a confusing interface that they've forced on to people.

Apple is doing AI right. It controls end to end everything it feels it needs to, but it partners with others to fill in where Apple doesn't want to play. It's an approach that will work and it's an approach that makes the most sense, to everyone but the markets.