Apple Intelligence Explained

Apple Intelligence was revealed during the WWDC 2024 event in June 2024. It was a monumental change to the entire ecosystem because, for the first time, Apple has publically jumped on the Artificial Intelligence bandwagon.

Let’s explore what is Apple Intelligence and what makes it different than other offerings by other AI-focused companies.

AI Features

Instead of a standalone product, Apple decided that AI would be a new feature for their ecosystem. In other words, Apple Intelligence will enhance the experience of using their products. Apple will cut your work from hours to mere seconds.

Writing Tools

Writing Tools will help generate, summarize, or expand the text that you selected. Wanted to write a professional email but don’t have the time, just key in the points, or even tell Siri to craft it for you. If the Apple language model is not to your liking, you can use ChatGPT to do it. Apple plans to add more AI models in the future. Expect Google’s Gemini to join in the fun.

Swift Assist

In Github, a popular code repository website owned by Microsoft, there’s a paid feature called GitHub Copilot which you can tell a prompt what you want to build, and the GitHub Copilot will spew out the code that will make it work. It’s starts from a $10-a-month subscription per user but significantly higher for businesses that want more features.

Swift Assist, an AI assistant in Xcode, Apple’s official IDE that helps developers build apps does this same thing. However, there is no word about pricing and if all indications are true, this feature will be free. Of course, the GitHub Copilot can do a lot more than just spew out code, but the core feature in Copilot would be free in Swift Assist.

Image Generator

There’s now a free-standing window where you put keywords to a box, and that program will generate images or icons based on your input. It can generate new images based on your library of images, like pictures of your friends and such.

Siri Integration

One long-standing complaint about Siri, Apple’s digital assistant that is embedded in almost all of their devices, is Siri is not that good compared to their competitors like Google Assistant or Amazon Alexa despite Siri arriving on the digital assistant scene first.

Apple Intelligence intends to make Siri a whole lot better by understanding your context better. For example, Siri will be aware of your screen and understand when you say, “Send a message to Mom: I’m on my way” or “Copy the cat picture from Google Chrome and send it to Mom”.

Integration with ChatGPT

ChatGPT is an AI-enhanced chatbot by OpenAI, a company specializing in AI solutions which Microsoft has a significant stake in. Their breakout was they they released ChatGPT 3.5 which is a leap forward in what AI can do.

Under The Hood

Apple created its Large Language Model (LLM) that powers the various Apple Intelligence features. Not only that, they have a Small Language Model (SML) that is designed to run on-device like their best-selling iPhones. One of the great things about SLM is they are small so the response is faster and more importantly, private.

Yes, SML might be limited compared to their big brother LLM, but if you can make the SML good enough for everyday use, you don’t need an LLM.

Apple also created its own cloud LLM in case the SLM is not sufficient. Unlike other companies that rely on harvesting user data and such, Apple Private Cloud Compute is LLM with Apple characteristics, mainly ease of use and privacy.

Apple allows interaction with 3rd party language models like OpenAI’s ChatGPT and Apple claims that data sent to OpenAI will be anonymized to protect user’s privacy and Apple’s reputation. We expect any future integration with AI models will follow the same rules that Apple sets.

Impact on Future Apple Products

Apple Intelligence has given new motivations for Apple engineers and developers to increase the performance of their devices. The major aim of Apple Intelligence is to have a private, scalable AI model that can be run on devices with cloud backup. Emphasis on the on-device capabilities which can run without the Internet and far faster than online solutions.

Based on Ollama, an open-source large language model solution, an LLM with 70B parameters would require 32GB of memory while an SLM would need at least 8GB of RAM. Right now, the iPhone 15 Pro, the current high-end smartphone from Apple has 8GB of memory. With competitors pushing 16GB and even 18GB of memory, Apple can easily put more memory on the next iPhone Pro to leverage the SLM.

Why Apple’s Approach is Different

While the topic of AI has reached explosive growth for the past few years, Apple’s approach is markedly different from its competitors like OpenAI or Google.

For starters, instead of touting Apple Intelligence as a standalone product, Apple Intelligence is a feature of the ecosystem that Apple spent years building.

Secondly, Apple Intelligence features are available to any developers who want to integrate the features into their applications. Not only that, existing applications that are built on the Apple ecosystem can utilize the features without any modifications. So things like Writing Tools and Image generation are available from day one, provided the end users update their OSes to the latest version.

Conclusion

Apple has moved AI from a fad to a mainstream feature. The grandiose promises that AI companies are pushing might not come to fruition, but we know at least the basic features like summarizing the avalanche of information into news headlines will be here to stay.

Plug

Support this free website by visiting my Amazon affiliate links. Any purchase you make will give me a cut without any extra cost to you

Other Ecosystem Items

• Apple TV (Amazon) / Apple TV 4K (Amazon)
• AirTags (Amazon)
• HomePod Mini (Amazon)