Apple’s AI-focused M4 chip: what AI tasks can it handle?

Apple is the latest company to announce tech products aiming to capitalize on the rapidly expanding realm of AI.

The company recently announced its newest iPad Pro line with OLED screens. It was the first time that a tablet made by Apple came with the latest M-series processor. Previously, the debut of M-series chips was reserved for the MacBook line.

The company claims that the M4 delivers up to 1.5x faster central processing unit (CPU) performance than the M2 in the previous iPad Pro. The latest Apple silicon has up to 10 cores – including six efficiency cores – and 10 Graphic processing unit (GPU) cores that are built upon the next-generation graphics architecture of the M3 family of chips.

Basically, it means that Apple has put a very powerful processor in a tablet, with the full potential of its capabilities being quite difficult to utilize.

Another notable feature is a Neural engine, which is the fastest made by the company, capable of up to 38 trillion operations per second (TOPS).

According to Apple, a Neural engine that, together with next-generation machine learning accelerators in the CPU, GPU, and higher-bandwidth unified memory, “makes M4 an outrageously powerful chip for AI.”

So, what AI tasks can the new iPad Pro handle?

The primary function of a Neural processing unit, or, as Apple calls it, a Neural engine, is to handle AI and machine learning operations right on the device instead of sending data to be processed on the cloud, thus enabling faster and more secure performance.

Specifically, Apple claims that with the M4, it will be possible to use audio or video text transcription using Live Caption or identify objects in video and photos using Visual Look Up in real-time.

The Neural engine will also be implemented in tasks such as isolating a subject from its background throughout a 4K video in Final Cut Pro with one tap. The M4 iPad Pro can also automatically create musical notation in real-time in StaffPad by listening to someone play the piano, the company claims.

These workloads can be done better with a Neural engine because it enhances the efficiency of GPUs and CPUs by handling various tasks, including blurring in video calls or object detection in video or photo editing. This allows a GPU to dedicate its resources to more demanding operations.

Apple claims that this is the fastest Neural processing unit of any AI PC today. While this statement is true right now, in the next two weeks, we should see several new devices with Qualcomm’s Snapdragon X platforms from Microsoft and ASUS. Qualcomm is promising 45 trillion TOPS.

While AI devices are currently the hottest trend, it’s also important to note a lot of AI hardware has limited capabilities. Why? Because there isn’t really that much software yet that can allow the devices to reach their potential. However, this should change in the future since many companies are working to embed AI capabilities in their applications.