As the world’s biggest telecommunication equipment celebrates it’s 30th anniversary, Huawei finds itself at a sort of crossroads as it plans its future in the mobile market. In 2016 Huawei solidified its position as the third biggest handphone brand by shipment behind Samsung and Apple, and in the first half this year they’ve already shipped over 70 million handsets. Yet a significant percentage of Huawei’s mobiles are sold in its home market China, where Huawei is facing increased competition against other Chinese handset makers such as Oppo, Vivo and Redmi.
To take a step ahead of its competition and truly evolve into a global smartphone brand, Huawei has begun to change the way it innovates, and this was made loud and clear at iFA 2017 when, during the brand’s first ever Keynote Presentation at iFA, Huawei’s CEO Richard Yu announced the incorporation of more powerful artificial intelligence (AI) with the launch of Huawei’s latest bespoke mobile processor, the Kirin 970.
The Kirin 970 is powered by an 8-core CPU and new generation 12-core GPU, and features a dedicated Neural Processing Unit. With native AI processing complemented by Cloud AI the Kirin 970 is able to perform AI computing task far faster and with much more efficiency – Huawei stating that in a benchmark image recognition test the Kirin 970 processed 2,000 images per minute!
The Kirin 970 will be available in the upcoming Huawei Mate 10 and Mate 10 Pro, which will be unveiled in mid-October 2017.
A few weeks prior to Huawei’s iFA Keynote address, we had a chance to interview Mr. Bruce Lee, Vice President, Handset Business, Huawei Global Consumer Business Group and he answered some questions pertaining to why Huawei was researching On-Device AI. Note this was before the Huawei iFA keynote so some specifications were never not mentioned.
VRZone: Huawei already has incorporated AI into its existing smartphones. When you say that Huawei is to further invest/focus more in AI development, what can we expect Huawei to push out in the future?
In the past, smartphones were limited by their processing capabilities, computing capabilities and battery life, there are very limited AI functionality that can be built in the smartphones. Furthermore, many of the traditional phones that offer AI incorporation offer the service through cloud computing.
We feel that in the future, more AI capabilities should be incorporated into the smartphone itself. However, we are limited by the CPU and GPU capabilities of traditional smartphones, so a lot of AI and machine learning functionalities are currently not suitable to be incorporated in the smartphones.
Hence, we feel that we need to work on better AI integration to achieve better computing, as well as better hardware for processing. There are a lot of advantages of local AI computing vs cloud computing. There is higher latency involved using cloud computing to push AI functionality to the phone, because of local data transmission to and from the cloud.
This process of real-time data transmission to and from the cloud is also restricted by our current broadband network speeds. Consumers also have privacy concerns of using cloud computing for AI functionality. With strong AI capabilities built into the handset directly, we can solve all of these issues.
VRZone: Is there any possibility to share numbers/company data in terms of productivity increases (cloud computing vs on-chip computing)
It’s a bit hard to compare due to the large differentials in power consumption between cloud-computing and on-chip computing. In the cloud, we can have all the different computers and processors to achieve all the different functions, but this is limited in the chip. In future, AI in the phone and AI in the cloud will work together in a synergistic way to provide all of the AI functionality to the phone.
VRZone: Can you share some real-life scenarios on how people can benefit from devices with AI?
For instance, if you book an air ticket on an AI-enabled phone in China, the system will remind you on the departure date, book a taxi for you, tell you the weather at the destination, friends at the destination and places you can visit. If you booked a high-speed train ticket on the smartphone, the system can also remind you when you’re reaching the destination, as well as remind you on your exact location of your car at the parking lot.
VRZone: Other than travel, are there any other examples?
In future, smartphones will be able to hear ambient sounds and noises, learn about the user, and become a personal assistant tool.
VRZone: How will Huawei tackle the issues and limitations that Samsung’s Bixby personal assistant, since they are developing AI as a personal assistant?
As far as AI is concerned, we don’t think any single manufacturer can fulfil all of the users’ needs. Huawei will provide a platform, with the hardware and computing capabilities, and will open the platform to our partners to develop apps to fulfil the needs of our end users.