Huawei today held a session where it talked about the possibilities of Artificial Intelligence in smartphones and how the company is looking to make the best use of that technology. The session saw James Lu, Director of AI product management at Huawei Consumer Business Group talking about the Chinese company’s plans for its future smartphones and the use of AI in them.
Last year, Huawei introduced the first AI processor called the Kirin 970 that featured a Neural Processing Unit(NPU). With it came upgrades to the ability to zoom further while not losing the clarity of the image, better integration of cloud-based and device based AI making the experience better.
And now with the technology in hand to recognize and differentiate objects of hardware created by Huawei and using Tensorflow and Machine Learning, Huawei has come up with its own APIs that govern the Huawei Mobile AI. According to Huawei, its Mobile AI will be using six engines that will be answering the challenges of decision making, self-learning, cognition and perception to work.
This engine is what gives the smartphone camera the ability to understand when the camera is too close to the object and that the user is trying to take a macro photo. With this, the company also plans to incorporate the combination of on-device and cloud AI computing to give users the ability to use AI functions while online as well as offline. Huawei alleviated concerns that this might result in wastage of storage space on the device by saying that the learning will be done on the cloud by the resulting code will be stored on the device which takes a fraction of the space.
Lu explained that the company is looking forward to developing the AI to a point where the phone will be able to predict and suggest services that the user might need without any prompt. Though he did not go into details of how this might happen or what services these might be.
As for the security concerns of the storage of user behavior and trends data, Lu said that the data will be stored on the device itself and would not be transmitted to the cloud servers. He explains that the new processors are powerful enough to compute all this on its own without having to send it over to the cloud servers. This does raise the question of how the data will be computed without wasting internal storage of the device.
All this being said, Huawei also explained the importance of having a AI ecosystem for the furtherance of the technology. To this end, the company is all set to give developers access to the API of Huawei HiAI for free to use the technology to create AI apps. The challenges that the developers will face will be easily handled by the high powered processors. This will help Huawei create an ecosystem along with its partners.
Lu says that an example of this would be the popular image editing app Prisma that typically requires at least eight to ten seconds to transmit the data on to the servers and process it there and send it back. But Huawei will be implementing some of the filters on the P20 itself that will cut down the time in less than half by processing the image on the device itself. Huawei says that this will bring down the processing time to three to four seconds while the device is offline.
Asus Zenfone 5Z First Look
Additionally, the AI will be able to give details about the image to the software on the device itself like if it is in portrait mode or landscape, what objects are in the image, etc. This according to the company will make AI a more available technology for both the developers as well as the consumers.