This year’s main smartphone launches promote Artificial Intelligence as a life-changing feature.
AI has been present in mobile phones for a while now, powering voice assistant features like Siri or the Google Assistant, for example.
However, in the previous generation of phones, AI was cloud-based and required an internet connection to be accessed.
What is different about AI on mobile devices is that the new generation of smartphones will combine the cloud-based AI to built-in AI engines on the hardware. This novelty has been announced by tech giants such as Google, Apple and Huawei.
What is in-device AI and how does it work?
From what experts have promised, it seems like the difference of having a smartphone to an ‘artificially intelligent’ phone is like the difference of having a pet to a guide dog. While a pet will obey your commands, a guide dog will not only respond to your orders but lead the way and make decisions to what it believes is best for you.
Huawei’s Mate 10, released today, carries the brand’s own Kirin 970 processor. On top of the CPU and GPU, common to processors, Kirin 970 also features a Neural Processing Unit (NPU).
While CPUs and GPUs can make precise calculations, AI units, such as the NPU, work differently.
Like the human brain, AI won’t make overly complex and simultaneous calculations instantly. What it will do, through sparse processing, is recognize images, voices and language and process them like data.
This means that phones like the Mate 10 will be able to make decisions and optimise their performance based on what they have learnt from being used. In-device AI also promises a better integration of the system with hardware such as cameras, microphones and batteries.
This leads the way for a number of unique features and improvements.
What will in-device Artificial Intelligence do?
On the iPhone X, AI algorithms are behind the facial recognition system Face ID, the animated emojis Animoji and augmented reality apps.
On the Google Pixel 2, one of the AI-powered features is Now Playing. Using a database of 100,000 songs updated weekly and customised to location, the phone will tell you immediately what song is playing in the background by automatically pushing a notification on the home screen. No need to access similarly dedicated apps like Shazam.
By understanding language and images, AI will also boost Google features such as live translations and Google Lens; the google search you can trigger by taking a picture.
On the Mate 10, AI-vision cameras scan the surrounding environment to adjust the phone’s settings accordingly, optimising performance and user experience.
When taking a picture, the phone can also identify the object, person or situation being photographed. A correction filter is also applied, resulting in better quality images. The image recognition and enhancement are accurate to such a degree that they can distinguish a cat from a dog.
The AI engine is also able to compensate for low-resolution images. If the picture being viewed on your device is of a lower resolution, the AI-vision engine can predict the pixels that are missing. This allows a better full display and zooming in.
All three brands have also announced a better battery performance on the new generation phones. This is partly due to the in-device artificial intelligence. Huawei has stated that its AI will make the company’s new phones 50 times more energy efficient.
In-device AI saves power in two ways: Firstly, it allocates power to relevant functions. So, if the user isn’t using a certain feature, the phone will not allow it to consume as power unnecessarily. Secondly, in-device AI means the cloud-based AI won’t be needed as much, which saves energy.
Experts also expect in-device AI to open doors to the development of all kinds of innovative and intelligent mobile applications.
“We’re really excited to see what the new app ecosystem will have to offer, developing new apps that will privilege the new AI capabilities on the hardware,” said Christophe Coutelle, Huawei’s head of software marketing.
For instance, The Microsoft translator application has been tested and can be accelerated by up to 300% with an in-device AI engine.