So at that moment we felt that there was a good chance that a series of chain reactions would happen soon thereafter.
In order to really dig this proposition down, this time we try to dispel the common disassembly mode of 'See Product Product' in the technology media. Instead, we start from the origin of mobile AI chip to observe a series of changes And create as much as possible to capture what exactly Mobile AI brings to us and what may happen in the future.
In a couple of months in a few articles, we have experienced the discussion triggered by Kirin 970; the advent of related mobile AI architecture; Apple released the iPhone X and A11 chips; Huawei Mate 10 as the domestic mobile phone AI play come out; so that Samsung , Google, Qualcomm, MediaTek have released the mobile phone AI program. And we also interviewed the chip development, mobile phone development related practitioners, AI researchers, mobile application developers, and more ordinary users, have discussed the development , Product, hardware, user experience and other related issues.
Perhaps, today is the time to come up with a summary of this proposition: AI computing on the terminal, what has brought the phone? What in the end consumers pay for the AI? From the mobile terminal to the pan-AI platform, the phone has started Some profound change?
And the most crucial, how do we understand the Huawei and other manufacturers put forward the 'smart phone into a smart phone'? - The wisdom of mobile phones, what is a gimmick or is it true?
Visual first: chip allows user AI more than voice
We summed up the three major changes brought by Mobile AI, bear the brunt of AI has brought a new visual upgrade.
We know that an important branch of artificial intelligence, but also the most hot area now is machine vision, and machine vision is also all-encompassing, and many of which can be reflected in the phone, such as mobile phones to take pictures to identify flowers and trees, such as Say capture dynamic objects, such as video tracking beauty, for example, the exact fit AR, such as face recognition, these functions are closely related to our daily lives are inseparable from the AI algorithm support.
And carrying a chip with AI-specific processing capabilities, the biggest improvement is that some of these capabilities can no longer rely on the cloud computing, after all, the cloud computing process has delayed the response, bringing the experience is not good, and in a non-network environment Can not operate.More importantly, some of the ideas that could not be run on the mobile end, the possibility of deployment in the phone.
For example, the faceID function of the iPhone X and the 'AI Eye Recognition' brought by the Huawei Mate 10 are all the 'fit technologies' that machine vision technology brings with the mobile phone's own image production mode.
More importantly, the mobile AI brings changes to the mobile phone images, such as shooting (Mate 10 smart photography), special shooting (algorithm processing light), dynamic capture (algorithm capture), video (smart video, landscaping), live Beauty), faceID, image recognition, AR and even MR.As can be seen from the product features of AI handsets such as Mate 10, no part of the visual or video is that AI can not go in to blend in one The future of these capabilities around, there will be room for further escalation.This undoubtedly ease the homogenization of competition between mobile phones.
The use of AI blessings as a central platform for handling image data in our daily life also means that an important function of AI can meet with consumers and greatly improve the awkward situation of only voice interaction in the field of user AI
From the point of view of human physiology, after all, experience and use of voice interaction is rather weak, and the mobile phone AI allows users to see the truth, and also arouses the commercial possibility of the AI itself.
AI from the beginning of the image, has become the next AI terminal AI must pass through the road.
Terminal Computing and Sensing System: Human-Computer Interaction Try to Step Into 3.0
Terminal AI computing brings a second change to the phone, from the computing power of the terminal with the sensing system, may permanently rewrite the history of human-computer interaction.
In previous articles, we discussed the awkwardness of mobile phone learning user information, and users do not want to open their own information to the phone, of course, the biggest concern from the potential risk of leakage.And the arrival of mobile AI chip means that the user information The learning and understanding can be carried out in a totally unoccupied enclosed space, so it is clear that users will be able to get rid of the biggest worry.
In addition, the sensor system upgrade brought by AI also means that the way mobile phones collect and understand user information is becoming diversified and diversified.
The two add up, on the one hand the information is more secure, on the one hand more casual information can flow into the cellphone, perhaps indicating that the cellphone understands that the user will enter a new era.
In Mate 10, we can already see the habit of machine learning users and the output of personalized programs in several factors such as photographing, power saving, system cleaning, intelligent translation, etc. This mode may still be relatively elementary today, but at The next maturity, but it may bring huge changes to the entire mobile phone industry.
The human-computer interaction process of the mobile phone has gone through two versions: the key control panel of 1.0 and the touch-screen control panel of 2.0, and once the terminal AI learns user data and the system is mature, perhaps the human-computer interaction will come to 3.0 mode - Humans rarely control or control the phone in a natural way.
If one day, the phone no longer need to configure the touch screen, and obviously many things have changed.
The battle of developers has been started: AI imagination open source
A third change that mobile AI has brought to mobile phones is the unprecedented increase in the value and imagination of the developer's ecosystem.
We noticed that handset-side AI developers and the development ecosystem have been the strategic resources that have to be contested from the outset. Around the machine-learning platform docking, the ease of use of the mobile AI architecture, and even the migration costs between Android and iOS , Handset manufacturers from the beginning of the mobile AI era on the idea of combining AI developers and mobile developers to form a new ecological community.
The most representative, to be unicorn 970 launched HiAI artificial intelligence mobile heterogeneous computing architecture, and actively open its source is to pave the way for developers to enter the mobile AI threshold.
Apple and Android vendors are deeply connected with the developers, the significance lies in the mobile AI provides the basis for calculation, hardware structure and development architecture, as well as a small part of the demonstration of AI applications AI as a highly bionic nature of the difference One of the characteristics of the new technology and characteristic of the traditional computing model is that it can construct a variety of application scenarios through calculation and data foundation.
So some people say that the AI era is bound to be crowdsourcing entrepreneurship era, because similar human-machine learning or deep learning platform, two people developers even difficult to develop a completely similar application model.
Back to the phone, this theorem should still be established in the chip to the hardware to improve the platform, we may see will be completely different, weird mobile AI applications. Different from the traditional mobile applications in the connection point differences, mobile AI applications will begin to differentiate in the underlying technology to achieve a true kaleidoscope era.
On the other hand, want consumers to really begin to pay for the AI.More important is to take out the irreplaceable evidence of mobile AI, and the best evidence is the application of explosive class spike level.So in the next period of time, Mobile AI developers are bound to go up, and being able to make something delicious on prepared ingredients is the result of a combination of mobile platforms and developers.
Summary: The wisdom of the phone
From the above three points is not difficult to see that the mobile AI brings changes to the mobile phone is quite back-end and multiple.On the one hand, it has had a profound impact in several key areas, on the other hand, consumers are also very easy on the mobile terminal quickly Perceive the existence and value of AI.
When mobile phone perception, learning, decision-making capabilities continue to increase, the increasing number of applications, the status of mobile phones may be from the mobile terminal to the pan-AI platform transition .And from the industry status quo, all manufacturers relay layout AI terminal calculation should not be very Far-reaching thing.And the AI developer community mature, it will force mobile phone manufacturers to configure AI speed.
In other words, Mobile AI will have a high probability of becoming standard for the next handset and may even begin to spread to lower-end products, and AI may also take this opportunity to become a new value proponent for the handset market by introducing a new premium module. Or the mainstream manufacturers to a new starting line.
The war on the starting line has begun to start, the most significant example is that many foreign mainstream evaluation sites are in contrast iPhone X and Huawei Mate 10 AI similarities and differences, put Samsung Note 8 thrown behind the body. Mobile AI, you can That is a victory of a domestic brand landing beach landing.
Finally, we may be able to summarize what is the so-called 'smart phone'. Some people say that AI + smart phone is a smart phone, it is a bit too mysterious. This fact is not understood what the so-called wisdom in the end.
In my opinion, smart phones do not mean that mobile phones can do anything, but rather that using AI algorithm blessing and data processing, machine learning and self-growth capabilities, mobile phones will likely get closer to the human intelligence model Some ability.
Human intelligence is a very complex ecosystem with some unique factors that are difficult to simulate in the classical computing paradigm.Of course, in the short term, the mobile phone can not completely or even not be able to simulate most of the human intelligence and can simulate Just a few features. For example:
Wisdom is what he can not do
Human beings through their own wisdom to judge, identify, propose solutions, these can not be replaced by other capabilities in the mobile terminal is also true. Join AI, mobile phones will be able to solve some of the traditional hardware and software systems can not solve the problem, such as in complex Choose from the camera features, such as the system to understand and remember your voice commands. For the traditional can not, that is wisdom.
Wisdom is vastly different
Even twins, their wisdom is not the same as two people with only the same level of wisdom, there is absolutely no same wisdom.This is reflected in the phone AI, pointing to a sound development environment for AI, cell phone products will make the difference between , Personalized, systematic product features.Privacy of mobile phones should have different things, just like the human brain, which is the wonderful imagination brought by AI.
Wisdom is a flash of light
Human use of wisdom, mostly in the electro-optical flint to make a decision.If you need to go home check a few days and then come to the conclusion, then we generally say that he profound, rather than wisdom.Good machine is also true, it should not search, Screening and then give the answer, but should be in an instant to meet the complex needs of users, mobilize all the ability to solve one thing.Mobile AI into the terminal computing era means that the mobile phone instant solution begins to come .How to train this ability to The user base has more significance, but also depends on the follow-up continuous exploration.
The so-called smart phone, and not so mysterious.Maybe just let us use the phone has become a little different.But that is, always embrace a little different, human beings have a future at all.