For the smart phone market, which is now 'scratching', the full screen seems to have become the product of the last century. If the conference does not mention 'artificial intelligence' things, I'm sorry, the next wave of mobile phone shuffling , Maybe it will soon be 'second kill' by other vendors.
From the second half of 2017 onwards, almost all of the mobile products were released with the concept of AI. For example, Samsung's AI assistant Bixby can complete intelligent translation via camera, exchange rate conversion, etc. Huawei also released the first integrated NPU neural network. The chip of the processing unit, the later released Mate10, P20, Glory V10, Glory 10, and other products all incorporate the function of photographing and identifying objects. Glory 10 is able to perform deep learning on professional photographers' views, parameters, and composition experience. Even the low-key Apple added a 'bionic' gimmick to the chip, indicating that it has a neural network engine that counts 600 billion times per second, and the use of the word 'neural network' has also driven this year's wave of mobile phones. Conference PPT follow suit.
However, AI is more like a marketing gimmick at the moment, and the stereotyped technology rendering does not bring more good news to this market in the first quarter. According to the "Report on the Operation of the Domestic Mobile Phone Market in March 2018" issued by the China Institute of Information and Communication Technology. , Shipments of domestic branded handsets fell by 27.9% in the first quarter.
'AI is not a function, a module or a technology, but a complete system composed of chip algorithms, systems, etc. If a company wants to build a real AI capability system, hardware, smart systems and applications must not be Shao Ming, President of Glory Zhao Ming pointed out at the GMIC Global Mobile Internet Conference that there are a large number of 'follow-up' AIs in the industry at the moment, that is, packaging AI with no substantial increase in computing power, and hype AI with no functional landing, lack of ecology. Closed AI building.
New gold content geometry
Huang Yuxi, a research manager of mobile storage at the state-level consultancy, told the First Financial reporter on the 2nd that in view of the smart phone field, upstream manufacturers have included AI applications in chip development in recent years. It is believed that the application of AI will become more popular and the antennae will also become more common. It extends down to the mid-range mobile phone. However, if the brand wants to make a difference, it must invest huge funds in the construction of AI applications.
In his view, AI is divided into three levels, including the first layer of chips and hardware, hardware that fits AI algorithms for the core processing of AI, and the second layer to create AI intelligence systems to enhance self-learning awareness. Vision, there is a third layer, application ecology. At any one of these levels, you can enjoy the dividends of this industry development.
However, in reality, more AI handsets are still in the “following” marketing status on the market. For example, using older chip architectures, there is no essential upgrade, but they are 'packaged into' third-generation, dual-core and other AI chip concepts. The computational power is lagging far behind the industry level. For another example, AI's name is given to the shooting function, and the promotion is greater than the actual experience. However, the actual photographing effect is not good, and even some of them are only identified on the PPT packaging scene. Classification, there is no manifestation of the phone function.
'Indeed, many of the AI mobile phone's emphasized capabilities are still mainly focused on image recognition and processing.' Huang Yuxi told reporters, such as opening the camera, if you point the camera at a person, then a preview will appear in the lower left corner of the screen A small icon for 'man'; if you aim at the plant, a small icon of the plant will appear. More precisely, this AI is still in the 'baby' state, which requires the user to send instructions to the AI on the phone to play a role. The implementation of the "passive reaction".
According to the current development of AI, most of them are based on single AI applications, such as voice recognition, smart environment sensing and photographing, etc. 'Yan Yuxi told reporters.
'But in reality, artificial intelligence will become a coprocessor of the human brain in the future, shortening the gap with experts. ' Zhao Ming told reporters that AI technology allows mobile phones to self-learn based on user habits, and then optimize, intelligent perception users Scenario: Predicting user behavior and then intelligently allocating resources. More intuitively speaking, with the application of AI in mobile phones, the mobile phone starts to actively observe you, understands you, the mobile phone turns out to be only a tool, but in the future it will become an outside brain, not Just another hardware state.
Can 'rescue' the stagflation market
With the intensive release of AI mobile phone products, a field AI mobile phone marketing war rushed.
However, no matter whether it is the open position of the person in charge of the mobile phone manufacturer or the data disclosed by the research institution, the anxiety and staleness of the market have not been eased in the first quarter. On the contrary, the survival of more small and medium mobile phone manufacturers has become more difficult.
Wu Qiang, vice president of OPPO, said that in the current T-shaped situation, the competition of the head enterprises is fiercer. In his view, first of all, one cannot make mistakes on their own. If they make mistakes, the scale will soon be eaten by other competitors. You can wait until the competitors make mistakes. , to get the other's market share.
The siege of head companies has put more small and medium-sized brands under pressure to continue to reduce their share. From the data rankings provided by the analysis agency Saino, it can be seen that the Chinese market has a marked polarization in the first quarter, with the top six vendors in the first quarter. The shipments are all in the tens of millions. Starting from the ranking seventh Meizu, the shipments of Jinli, Samsung, and Chili are all in the order of one million. This also means that the mobile phone manufacturers ranked outside the ten, shipments Less than a million in the first quarter, with an average of 400,000 sales per month failing.
For small and medium-sized brands, the harsh winter of the market is an indisputable fact.
According to many analysts, it is still unclear whether AI can save the 'stagflation' of the mobile phone industry. However, it is undeniable that the combination of mobile phone and AI will be one of the future technological trends.
Huang Yuxi believes that if science and technology iterations always require cycles, mobile phones will enter the era of true AI, open up subversive changes, and find landing scenes need to take longer.
Yu Chengdong, the head of Huawei’s terminal, once called the chip that integrated the NPU artificial intelligence processing unit “nuclear weapons” in the chip. A research and development team with tens of thousands of people stood behind a mobile phone chip. All of them are masters and Ph.D. The NPU of the independent hardware processing unit, the NPU independently completes the special AI operation task, and improves the AI computing capability.
In contrast, Qualcomm uses the three major modules of the CPU, GPU, and DSP for AI operations, and analyzes and issues specific AI tasks through a neural processing engine (SNPE). MediaTek uses the 'dual-core APU'. The ability of multiple DSPs to improve the efficiency of image post-processing, and then do AI-related image processing algorithms.
"Like AI, chips are heavily invested, and it takes only 10 years for a single investment. It took ten years to build the chip that leads today in artificial intelligence." Zhao Ming told reporters that when it comes to full comparison and choice, the strong will Stronger, mobile phone brands will become more and more concentrated.