You may not know the ARM, but you certainly use technologies you create on a daily basis. The British company, controlled by Softbank, is responsible for the industry’s most popular mobile processor architecture design – a model that is followed by several manufacturers, including Qualcomm and even Apple.
On Tuesday, the ARM revealed the creation of two new chip designs designed to process artificial intelligence algorithms. The initiative has been named Project Trillium and should help make more devices have native artificial intelligence in the near future.
One of the designs is called ARM Machine Learning (ML) and serves as a guide for companies that want to assemble processors prepared to handle applications that use machine learning. This includes real-time translation, facial recognition, and similar software.
The second drawing is called ARM Objection Detection (OD), and is used to process visual data such as photos and videos, recognizing objects and people. The idea is that the design be used by manufacturers in the creation of chips for security systems, from smartphones to cameras and PCs.
As with its processor architecture, ARM will not manufacture the chips, but will license the patents so other companies can build them. Based on a standardized architecture, more phones are expected to come with artificial intelligence chips in the coming years.
The advantage of this technology can be felt both on the industry side and the user side, as The Verge explains. Currently, artificial intelligence applications – such as Google Photos image recognition or the Siri virtual assistant – require connection to external servers.
To recognize your voice command, Siri relies on Apple computers connected to it over the internet, not the iPhone itself in your hand. That way, the answer lasts a bit longer, consumes bandwidth, takes up Apple’s servers and can still be intercepted midway.
With a native artificial intelligence chip, these problems can be reduced. Siri could, in theory, recognize your voice command locally, without Apple’s servers, ensuring a faster response and safer interaction with no risk of interception.
Some companies have been adapting chips with ARM architecture to accommodate local processing of artificial intelligence, such as China’s Huawei. But with the new designs of the British company, it will be easier for the entire industry to make this transition in a practical and standardized way.