Title Picture | Visual China
Consumers have been tired of the mobile phone market for a long time, especially in the past two years.
According to a report by research agency Canalys, global smartphone shipments will decline by 4% year-on-year in 2023, and the Chinese market will show a similar trend, with shipments falling by 5% year-on-year. Behind this, in addition to the changes in the overall economic situation, consumers have become more sensitive to price and pay more attention to cost performance, and the increasing convergence of smartphone functions has also caused consumers to have aesthetic fatigue. In the past, those innovative functions around music, photography, style design and other subdivisions, after consumers actually used them for a period of time, found that the difference between most mobile phones in basic functions was not significant, which further weakened consumers' desire to buy.
So, a question arises: how can mobile phone manufacturers break the silence of the market and realize the revival of the mobile phone market?
In response to the above problems, the current mobile phone market presents two different development paths.
First, choose cross-border empowerment and apply the technology and user base in the field of smartphones to other terminal scenarios, such as the automotive field. Huawei released a new strategy for smart mobility solutions to deepen its efforts in the automotive industry, while Xiaomi has fully entered the field of new energy vehicles, attracting market attention. However, this kind of horizontal empowerment still stays at the level of scenario application, and the challenges involving technology and supply chain should not be underestimated. At present, it seems that the cross-border car manufacturing of mobile phone manufacturers is more about business expansion than the transformation of core business.
Second, finding breakthrough technologies as a driving force to drive the development of other technologies and reshape the industrial ecosystem is another more challenging path. Generative AI, represented by ChatGPT, has made a major breakthrough in 2022, and mobile phone manufacturers seem to see hope for market revival. From the second half of 2023, Google, Samsung, Honor, OPPO and other manufacturers have built large models into mobile phones, and initially completed the deployment of AI in mobile phone terminals; Apple also abandoned its ten-year car manufacturing plan at the beginning of this year and turned to the AI field, planning to join hands with Open AI, Baidu and other companies to create a new situation in the field of mobile phone AI.
Although this road is full of challenges and requires long-term basic research and industrial practice, once AI mobile phones are successfully integrated into production and life scenarios, it will bring great benefits to society.
Taking the development path of the Internet era as an analogy, in fact, it is the integration of advanced mobile communication technology, which has successively driven the development of display, cloud computing, big data, blockchain and other related technologies, and finally constitutes the ecology of the entire Internet industry today, improving the exponential growth of the economy and the quality of human life.
In today's era, AI technology, which was once out of reach and full of mysteries, has become a "conversational application" that can be used on a daily basis. Whether AI capabilities can be integrated with mobile phones to create a combination of the two swords of the information revolution is a huge business opportunity, and AI mobile phones are expected to become a key factor in opening the door to a new era.
According to IDC's forecast, global shipments of next-generation AI mobile phones will reach 170 million units in 2024, accounting for about 15% of total smartphone shipments, a significant increase of about 233% from about 51 million units shipped in 2023. In addition, with the development of new chips and the further expansion of application scenarios, the penetration rate of AI mobile phones will continue to rise rapidly.
Is the general direction right?
As early as 2016, Honor launched the first generation of Magic mobile phones with the concept of AI, which put forward the concept of adding intelligence engine (Magic Live) to mobile phones, which can intelligently perceive the user's use environment and provide more suitable auxiliary services, establishing the direction of Honor redefining human-computer interaction with AI, bringing many revolutionary A experiences. By the time Apple launched the A11 chip in 2017, the industry had actually gradually found a breakthrough technology direction for the future of mobile phones - Al.
At that time, mobile phone manufacturers mainly applied AI to application optimization and chip improvement. Until now, most manufacturers have adhered to this strategy, although they have successively launched their own large models or grafted the general network large model into mobile phones: OPPO Launched a three-level large model deployment scheme (Titan super large model, Turbo large model, Tiny light model); vivo has also built its own blue heart large model matrix, for different scene applications, with different parameters (billion-100 billion) large models; Google, Samsung and other manufacturers also take the route of focusing on application function experience, advocating lightweight models.
It is not difficult to see that the strategic operation of most manufacturers in the market makes people feel as if they can be called AI mobile phones by installing large model technology on mobile phones and providing generative AI services.
However, Zhao Ming, CEO of Honor, recently threw out a "counter-common sense" point of view to consumers and the industry, and bluntly stated that "mobile phone AI is not equal to AI mobile phone".
In his opinion, the integration of AI technology and experience on mobile phones, mobile phones provide platforms and algorithms, which belong to the application of mobile phone AI. But AI phones mean using AI to reconstruct and enable mobile applications and operating systems, including all aspects of consumer experience and business logic. "AI is the foundation, and it is ubiquitous on mobile phones, which can be called AI mobile phones. ”
Zhao Ming's point of view is obviously "unconventional" enough, but at least the general direction of the mobile phone market to develop AI mobile phones has been clear, but the definition of the basic concept of AI mobile phones, product logic and long-term layout are still relatively vague.
Steve Jobs showed a coordinate diagram at the launch of the original iPhone to define the ideal form of a real smartphone, with the horizontal axis representing ease of use and the vertical axis representing intelligence, and pointed out that the iPhone was the embodiment of this ideal.
With the passage of time, in the face of AI mobile phones, we can still use the two coordinates (ease of use + intelligence) left by Jobs 17 years ago as a reference, and make new judgments on some issues that have not yet been clarified in the current environment of AI development, so as to correct the industry.
Ease of Use Coordinates: Is there a tool for digestion?
From iOS, Android to HarmonyOS, the operating system has been optimizing the adaptation of the hardware ecology and the Internet ecology, and continues to evolve by optimizing interface design, improving system performance, and enhancing security to improve user service experience. However, in the era of AI, traditional improvements cannot meet consumer needs, and the improvement of ease of use is limited. We also need to rethink the relationship between people and mobile phones.
Zhao Ming explained this problem with first principles, "From the perspective of first principles, what do you think the mobile phone should look like? Is the mobile phone commanding me, or is the mobile phone helping me? The mobile phone should know me best, understand me, know my needs, and thus serve me, this is the first cognition." ”
From the perspective of human-computer interaction, the current smart phone is still a machine-centered resource allocation and human-computer interface and natural interface interaction, which is based on the logic of iOS. In practice, people need to adapt to the operation of the mobile phone, for example, you need to enter, find, confirm, and execute.
But our ultimate goal is to serve people and become people's brains, not just as an intelligent tool, so that people can adapt to the tool, which is lifeless.
Therefore, a real AI mobile phone should have the ability to develop and evolve on its own, have a personalized and even personalized operating system, and eliminate its instrumental characteristics to the greatest extent. Such a mobile phone should truly understand the user, provide a personalized experience according to the needs of different users, and realize the so-called "thousands of people, thousands of faces".
How to build a personalized and intelligent operating system has become a common problem in the industry. Kai-Fu Lee, Chairman and CEO of Sinovation Works, said: "The development of artificial intelligence has entered a new stage, that is, the AI 2.0 era, which is not a specific product or technology, but a platform-level trend. ”
Just like electricity is a great invention, but it also needs to rely on the platform of the power grid to use electricity. Obviously, in this era, AI is electricity, and the operating system is the power grid, and we need to build a platform that is enough to undertake AI, which is a very good solution.
The platform-level trend that Kai-Fu Lee is optimistic about is actually not yet visible in the industry. However, we can get a glimpse of one or two from the relevant functions launched by terminal manufacturers.
There is an "ArbitraryDoor" function on Honor phones, which can automatically filter the services that users may need, so as to help users establish a direct and efficient connection with applications and services. This approach of breaking the traditional cross-application flow steps is logically the so-called platform-level AI capability.
First of all, it must have an AI technology framework that can support this function from the bottom of the system, and secondly, the AI it has must have evolutionary characteristics, that is, it must continue to remember and learn. In this way, the system can accurately identify the user's intent, efficiently schedule system services, and provide an intelligent experience of "guess what you think, understand what you need", which is completely different from the blank jump when we are usually faced with hundreds or thousands of apps on our mobile phones.
This kind of human-computer interaction under intent recognition is very consistent with "first principles". The fundamental logic is that mobile phones have been designed to serve people since the day they were born, and the iteration of panel, CPU, battery, photography and other technologies has always been to provide users with a better experience.
This can also explain "how to meet the ease of use of smart devices in the AI era?", and the answer is to be smart enough to understand you. It learns user habits and assists users in making decisions, with the ultimate goal of providing additional satisfaction in connecting human needs and terminal functionality.
This human-computer interaction logic based on intent recognition undoubtedly provides a clearer and more direct user link path for application developers, which helps them better expand the functionality of their applications, and also increases the reach rate of high-quality applications and Internet services.
Intelligent Coordinates: Do you think actively?
As mentioned earlier, the application of AI in mobile terminals is still in the exploratory stage, and some manufacturers have achieved rapid optimization of specific scenarios by integrating a variety of independent generative AI capabilities, such as intelligent translation and retouching, but the application scenarios are limited and are still in the stage of "we tell the mobile phone what to do".
At this point, the large model agent becomes the new focus, which can understand intentions, decompose tasks, and perceive scene changes and user habits at the operating system level like a human assistant, so as to undertake complex tasks such as arranging your trips, planning trips, organizing parties, etc.
More importantly, the agent does not target a certain type of problem or a single scenario, but uses operating system-level resources to perceive changes in the scene, learn user habits, think actively, and make decisions to call all the capabilities of the system. In this way, the large model can take on complex tasks, and the interactive experience is upgraded.
Apple's possible launch of generative AI technology at the upcoming WWDC2024 conference comes after a paper titled "Ferret-UI: Mobile UI Understanding Based on Multimodal Large Language Models" revealed some news.
According to the article, Ferret-UI can locate and recognize images and UIs on the one hand, and users can have more in-depth conversations with large models based on specific areas on the mobile phone screen. On the other hand, it can also recognize speech, text, etc., including detailed description, perceptual/interactive dialogue, and functional reasoning. What's more, it is embedded in the iOS system, not only for text recognition, but also for recognition based on the pages and images displayed on the current mobile phone screen, and then interact with the user.
In other words, Apple plans to launch an AI iPhone in a way that reimagines the system. This coincides with the platform-level AI strategy proposed by Honor in 2021 in terms of technical route. The "arbitrary door" function mentioned above is implemented in the Honor MagicOS 8.0 system. And the reason why it can be done is due to the "new core" of platform-level AI.
According to Honor, platform-level AI manages human-related factors, and MagicOS can accurately identify user intentions, efficiently schedule system services, and bring users an unprecedented intelligent experience. Translated, in order to make your phone understand you better, Honor has upgraded the operating system with AI, which can guess your thoughts before you do a certain operation, and then help you complete it faster.
And the reason why Glory did this is because the route was formulated early. In 2016, Honor introduced the concept of AI on the mobile phone, which basically determined that the future industry trend must be AI mobile phones;
To put it simply, although Honor has planned a four-layer architecture for its platform-level AI strategy, which are AI enabling cross-system integration, AI reconstructing the operating system, AI application on the device side, and AI device-cloud collaboration. But from 2021 onwards, it has done one or two layers first, and it has been done "secretly". It's not that the glory is subjective, it's mainly because these two layers belong to the foundation, and the external perception itself is weak. It's not like other manufacturers that start from the third or fourth floor, and it can be seen at a glance from a long distance.
But the simple truth tells us that if the foundation is not strong, the earth will shake the mountains. Only when the foundation is stable can the superstructure be continuously improved, and platform-level AI capabilities must be innovated from the operating system to achieve the best experience. Once it is done, it is a barrier that is difficult for others to easily overcome.
Of course, Honor is not without third-layer AI applications, Zhao Ming revealed in an interview that this year's goal is to focus on the third and fourth layers. And on the fourth layer, Honor has officially announced the 100 model ecology, that is, to provide an entrance for the 100 models, so that the cloud-side general model and the various industry vertical domain models can be integrated into the magic OS through shelfization. Among them, Baidu, AutoNavi and Air Travel have all joined.
epilogue
We once translated smartphone as Smartphone, and today's AI mobile phone, its Chinese name is artificial intelligence mobile phone, compared with the previous one, its name has added the word "artificial". We expect it to provide a deeper understanding of our intentions and more intelligent services.
Of course, it is not easy to meet human expectations for AI mobile phones, because it is not a simple hardware stacking or software upgrade, but requires manufacturers to have extraordinary insight and decisiveness before they can see the right route. In addition, manufacturers need to have enough understanding to implement technology with service as the first element in product change. Finally, it is the manufacturer's own strategic ability, which is solid and stable in formulation and execution.
There is no need to doubt the future of AI phones, just as there is no need to doubt that the future world will be AI. Moreover, to make another bold prediction, the AI carrier we can think of today is mobile phones, and there may be more carriers and forms in the future, in which the imagination space is not a communication tool of several thousand yuan, and every terminal manufacturer has more possibilities. And every technological innovation will lead to a huge change in the market, and ultimately benefit every consumer.
This content is the author's independent view and does not represent the position of Tiger Sniff. May not be reproduced without permission, please contact [email protected] for authorization
People who are changing and want to change the world are all in the Tiger Sniff APP