The AI Newsroom’s Post

In iOS 18, Apple plans to have all its AI features running on an on-device, fully offline model, Bloomberg recently reported. It’s tough to build a good multipurpose model even when you have a network of data centers and thousands of state-of-the-art GPUs — it’s drastically harder to do it with only the guts inside your smartphone. So Apple’s having to get creative. In a paper called “LLM in a flash: Efficient Large Language Model Inference with Limited Memory” (all these papers have really boring titles but are really interesting, I promise!), researchers devised a system for storing a model’s data, which is usually stored on your device’s RAM, on the SSD instead. “We have demonstrated the ability to run LLMs up to twice the size of available DRAM [on the SSD],” the researchers wrote, “achieving an acceleration in inference speed by 4-5x compared to traditional loading methods in CPU, and 20-25x in GPU.” By taking advantage of the most inexpensive and available storage on your device, they found, the models can run faster and more efficiently.

🍎 Despite using an iPhone and a MacBook Pro as my daily devices, I don't consider myself an Apple fanboy. Not owning an AppleTV, a HomePod, or a Vision Pro should be good proof of that! 🥱 And honestly, I've grown tired of Apple WWDC keynotes: being able to edit a message is not keynote-level content in my book. Over recent years, I've felt that WWDC was about making a big noise around software features that, for many other hardware/software companies, would only be worth mentioning in release notes. 💡 But the WWDC24 keynote (yesterday) was on another level: a masterclass in "how to craft products using technology." After several introductions of Machine Learning (the amazing iPad calculator app is one of them), came the "one more thing" moment we were expecting: AI as in Apple Intelligence. What was demonstrated is another level and really got me excited. Although I built myself an open-source AI-based desktop assistant (https://lnkd.in/giy2RqTS) that my family uses, the level of what was shown yesterday is opening the doors to a "Her" movie experience. If the new Siri (https://lnkd.in/gXPX9A5b) is really as good as it was shown yesterday, then I believe that apps will soon become a thing of the past and the only UI is going to be voice-based. Of course, all of this still has to be taken with a grain of salt, considering how bad Siri looks now versus when it was initially released... 🕵️ Data privacy was big on Apple's agenda. It has been in their playbook for a very long time, but what they showed yesterday lets me think that it was really a first-level concern in Apple Intelligence design. From running models on-device, requiring user-consent to use cloud models, and providing the source code of their servers to security specialists, they pushed the envelope! If you want to know more about the nitty-gritty technical details, including how they handle privacy, this video is a must-watch: https://lnkd.in/gXHPTFcH. 🍿 I have not been so impatient to install iOS and macOS betas in years. Not only to "play" with it but to actually use it! And if the rumor is right, my iPhone 14 will not have access to all the new goodies, but I will happily upgrade to an iPhone 16 in September! 💬 What did you think of those announcements? Are you as excited as I am?

Tim Cook hints at major AI advancements coming to iOS 18 and macOS 15, with Apple's generative AI set to outperform rivals. Cook emphasized the unique integration of Apple’s hardware, software, and services, enhanced by Apple silicon and a focus on privacy.🤳🏼🍎✨ #apple #AI #WWDC #generativeAI #genAI #appleAI #artificialintelligence #techinnovation #ainews #technews

Geekbench AI evaluates how fast your device can do machine learning and AI tasks.

It is "patently absurd that Apple isn’t smart enough to make their own #AI" said Elon Musk last week -- attacking Cupertino's #ChatGPT integration as part of its pending Apple Intelligence platform. Musk, of course, was inaccurate. As part of the release #Apple will be using a homegrown 3B parameter on-device model as well as a larger cloud-based one that makes use of tools like its own "Talaria" for optimisation. What you need to know 👉 https://lnkd.in/egbSFj35 #generativeai #ios #macos #llms

Here we are. Apple's time in GenAI. Yet this does not impress me much: simply, they don't have their own GenAI model yet, so they choose to team up with the best in class as of today, OpenAI. They did this before, for example, when they jumped on Intel's bandwagon while waiting to design their own CPU architecture. I think in one year or so, they'll have their own LLM. Please share and comment on our news article on this topic if you like it. #llm #openAI #apple #chatGPT #chatbot #ios https://lnkd.in/dvMYJBTr

Embedded systems, devices with a specific function within a larger system, have become an integral part of our daily lives. From smartphones to industrial automation, these systems rely on specialized operating systems (OS) to manage their hardware and software components efficiently. Let's explore some of the most popular operating systems for embedded devices.

Microsoft announced Phi-3 as an open source. Given the small size of 3.8B, it performs well enough to be deployed on iPhone 14. So, I summarized the tech report of Phi-3. You can learn its performance, model architecture, pre and post training scheme, and datasets. Please go for it! https://lnkd.in/gkJ8XHu9 #microsoft #llm #sllm #ai #opensource #iphone

🍁MacBook Air with M3 is ready for AI workflows. No matter whether businesses are using intelligent features in macOS, or apps using cloud-based AI, on-device AI, or a hybrid model, every Mac with Apple Silicon is already an AI PC.  Learn more: http://ms.spr.ly/6047YaSof #InsightCanada #Apple #MacBookAir #AIReady

🍁MacBook Air with M3 is ready for AI workflows. No matter whether businesses are using intelligent features in macOS, or apps using cloud-based AI, on-device AI, or a hybrid model, every Mac with Apple Silicon is already an AI PC.  Learn more: http://ms.spr.ly/6047YFMNz #InsightCanada #Apple #MacBookAir #AIReady