Buzz
This is quite a bold statement, suggesting that Apple's AI processing power surpasses Core Ultra and Ryzen AI on AMD 8000 series processors for laptops. And this seems plausible considering that the latest NPU in Ryzen 8040 Series 'Hawk Point' has a power of 16 TOPS, whereas Intel Core Ultra has only 11 TOPS.
The NPU integrated in AMD processors is still weaker than the Neural Engine in Apple M3.Specifically, with Core Ultra, you get a maximum of 34 TOPS, with 11 TOPS from the NPU, 18 TOPS from the GPU, and the rest from the CPU. With Ryzen 8040 Series 'Hawk Point,' you get a maximum of 39 TOPS, with 16 TOPS from the NPU, and the rest from the GPU and CPU. So, indeed, in terms of NPU power alone, Apple is ahead with its dedicated 18 TOPS from the Neural Engine.
Neural Engine is leading the game.. However, when it comes to laptops in general, the ultimate bosses must undoubtedly be those equipped with RTX 40 series GPUs, boasting processing power for AI tasks somewhere in the hundreds of TOPS.
In terms of hardware,
MacBook Air M3 is equipped with a faster and more efficient 16-core Neural Engine. Apple also states that the Neural Engine will be integrated with accelerators in the CPU and GPU to enhance machine learning capabilities on the device. Apple has been incorporating Neural Engines into their computers since the MacBook Air M1 in 2020.What makes AI on MacBook Air M3 superior to AI on Windows?
In reality, macOS is applying a lot of machine learning; it's a branch of AI itself. macOS can learn user habits, optimize battery life, feature intelligent photo organization and recognition, allowing you to directly copy text content from images effortlessly, and many other features supported by the Neural Engine.
What can AI on MacBook do?This is understandable as Apple has been developing the Neural Engine for quite some time. The Apple A11 on the iPhone 8 and 8 Plus in 2017 were the first processors to have a Neural Engine. iOS and macOS share many features, all developed by Apple. With 7 years on the market and several years of prior research, Apple obviously has clear plans and has software features tailored to its hardware.
The Windows world is different; AMD only initiated the NPU last year, and the software still cannot keep up as most Windows features that utilize the NPU are... somewhat redundant. And even if Microsoft releases a Windows version utilizing the NPU later this year or next year, it's probably difficult to catch up with Apple since they've been using the NPU for a long time already.
AI in Windows is just starting out, useful features all run on the cloudSo, for now, hardware strength comparison aside, it's true that in terms of software, macOS has surpassed Windows in utilizing the NPU for AI tasks.
But when it comes to using third-party software or training AI, it's a different story. For AI training, the NPU is still not suitable; currently, NVIDIA GPUs are the best choice that cannot be replaced. And as for software that can use AI, macOS offers much more variety than Windows. Because Apple has equipped Neural Engine for Macs since 2020, there are already many AI applications and software for Mac.
What can AI do on MacBook Air M3?
There are currently numerous AI software applications running on Mac. For instance, Luminar Neo app can enhance image sharpness with astonishing before-and-after results through its Supersharp AI feature. It can also minimize or eliminate motion blur. Other AI capabilities include easy object removal, background extension, and more.Another interesting Mac AI app is FreeChat, a macOS app enabling you to run multiple AI models and chat with them like ChatGPT or Gemini directly on your Mac. Note that these apps and Mac AI apps run on M2 and M1 chips but perform faster on M3.
AI on MacBook Air M3 can achieve quite a lotOther Mac AI apps include Goodnotes 6, providing direct AI math support within the app, Pixelmator Pro for image enhancement, and CapCut for removing background noise from your video. Additionally, you can run cloud-based AI apps like Microsoft Copilot in Word and PowerPoint, Canva, and Adobe Firefly. However, local software still offers certain advantages.Apple also quotes some specific performance standards for AI tasks. For example, the company states that enhancing images in Photomator's Super Resolution feature is up to 40% faster on M3 compared to M1. Pixelmator Pro is an app that allows for impressive background removal, leveraging the AI power of the machine.Below are some AI software applications that can run well on Apple Silicon Macs in general and MacBook Air M3 in particular.Pixelmator Pro
Pixelmator ProPixelmator Pro is a reputable image editing application for Mac, offering a set of intelligent, AI-powered tools, including enhancement and noise reduction tools for images.Pixelmator Pro is priced at 49.99 USD, but you'll get photo editing capabilities, vector design tools. It runs natively on Apple Silicon and utilizes Apple's Core ML for machine learning applications.Luminar Neo
Previously, Luminar Neo wouldn't make it to this list because most of the AI processing occurred in the cloud, but with the latest release, it brings local machine learning capabilities on Apple Silicon.
Luminar NeoThe price depends on the duration you want access, ranging from 12 USD per month to 80 USD per year. You can also purchase a lifetime license for 199 USD. AI features include sky replacement, enhancement, and Supersharp designed to make even the blurriest photos sharp.Pictura
Pictura also utilizes AI to enhance images. It comes with two AI models, both running locally on Apple Silicon and capable of quadrupling the resolution of low-resolution images.
PicturaCurrently, image enhancement features are free; if you want to save them, a lifetime subscription is 35 USD. It has models for photos and artworks or animations. Because it runs locally, it processes images faster than uploading to the cloud.Diffusion Bee
Diffusion BeeDiffusion Bee is a free macOS application that allows you to run Stable Diffusion models to generate images on your MacBook. It runs offline, and all data is stored locally. This free app enables you to convert text into images, images into images, and enhance image quality.Ollama
Ollama is a project implementing multiple open-source large language models including Meta's Llama 2, Mistral's Mixtral, and Google's Gemma.
OllamaAll these models run locally on your computer, and with the addition of a web interface, you can converse with them just like you do with ChatGPT or Gemini. This app is also free.FreeChat
FreeChatFreeChat is a native macOS app that allows you to run multiple AI models and chat with them just like you do with ChatGPT or Gemini.Unlike Ollama, which requires using Terminal for installation, FreeChat is a standalone, self-contained app available on the App Store.Arc Browser
Arc Browser is a modern web browser built for macOS utilizing AI for advanced customization features. It can learn your browsing habits and needs, adjusting viewing modes and screens to fit each content.
Arc BrowserIts AI usage is most evident in tab management and downloads. For example, it can automatically rename a tab to better reflect its content rather than just using the title determined by the website. It does something similar for downloaded files, renaming them to reflect the content/source of the file rather than any name given upon upload.Overall, AI on macOS Apple Silicon in general and MacBook Air M3 in particular is truly useful for users and clearly at the current time, it outperforms AI on Windows PCs, except of course for computers using powerful NVIDIA GPUs enabling running and training complex AI models.
See more:- What is an AI PC? Do you really need an AI PC?
- See more articles in the Market category
If you're interested in Apple Silicon Macs, check out these great deals now:
