Entering 2024, the application of AI large models will explode in all directions. The popularity of "Phantom Beast Palu" has allowed the industry to see the productivity changes brought by AI models. The emergence of SORA has made the world exclaim that the boundary between virtual and reality has been blurred.
But if you want more people to truly perceive the beauty of AI models, these are not enough, because the vast majority of users must be inseparable from a stage when they encounter AI, that is, mobile phones.
On Chinese New Year's Eve, Liu Zuohu, senior vice president and chief product officer of OPPO, announced that OPPO has entered the era of AI mobile phones. On February 20, OPPO held an AI strategy conference in Shenzhen, sharing the four capabilities and characteristics of the new generation of AI phones, and looking forward to the trend of AI-driven full-stack innovation and ecosystem reconstruction of mobile phones.
It can be seen that the starting gun for the comprehensive AI of the mobile phone industry in 2024 has been fired.
Prior to this, Samsung Galaxy S24 shouted the slogan of AI for All, and OPPO Find X7 became the world's first mobile phone to land a large model of 7 billion parameters, at the same time, Huawei, Xiaomi, Honor, Vivo and other manufacturers have introduced large models into mobile phones, and Apple will also introduce a large number of AI capabilities in June this year.
What does the opportunity of AI large model mean for mobile phones? How can mobile phone manufacturers seize this biggest opportunity in 2024? How to build a long-term AI development strategy?
In this article, we try to answer these questions by interpreting the AI starting situation of Samsung, OPPO and other manufacturers. Perhaps after deducing the answer together, we can reach such a consensus: daring to do AI and being able to do AI well is the most important juvenile spirit in today's mobile phone industry.
mobile phone, arriving in a new wilderness
In order to understand the new opportunity of mobile phone + AI large model, we must understand a premise: the ability of mobile phones to accommodate AI large models is not a straight track, but an open wilderness.
With the development of AI technology today, it can bring capabilities such as machine vision and natural language processing in terms of modality, and can be divided into decision-making AI and generative AI in terms of application purpose, so the integration of AI large models and mobile phones is also very complex. AI technology can be integrated into an application, into the software and hardware capabilities of the mobile phone itself, or as a system-level interaction solution for the mobile phone.
In the past few years, mobile phone manufacturers have paid the most attention to the image track. However, the image track is relatively more single-threaded, the standards are relatively unified, and the results are relatively clear. The difference between the AI model and the mobile phone is that its combination with the mobile phone has great openness and diversity.
This opportunity will lead the mobile phone market and the entire industrial chain, and become the biggest opportunity for the mobile phone industry in 2024. At the "OPPO AI Strategy Conference" AI mobile phone industry roundtable forum, representatives from all parties in the industry shared their views on AI mobile phones.
Fok Jinjie, president of IDC China, proposed that according to IDC**, there will be more than 1500 million AI mobile phones have entered the market, and the mobile phone industry has ushered in a new blue ocean space. The impact of this opportunity will run from the application side to the basic hardware of the mobile phone. Shen Bowen, head of product architecture at Feishu, shared the real changes that AI technology has brought to Feishu. The upgrade of application capabilities has led consumers to choose AI mobile phones, and mobile phone manufacturers are also required to do a good job in AI mobile phones. Chen Yiqiang, deputy general manager of MediaTek's wireless communication division, proposed that in the face of the opportunity of AI mobile phones, MediaTek will cooperate with OPPO to flexibly use AI capabilities according to user scenarios and user needs, so that AI can provide the best service to just the right extent and under the load of the system.
Under the common expectation of the industry, OPPO hopes to raise the banner and become a promoter of AI mobile phones.
Liu Zuohu believes that OPPO, as a contributor, hopes to bring the entire industry and the entire ecosystem into the era of AI mobile phones. Because AI mobile phones are an ecosystem, it is difficult to achieve them if they rely on one company and one brand alone, so OPPO calls on everyone to promote the development of AI mobile phones and define AI mobile phones together.
So, what exactly is an AI phone within the visible range at this stage? To this end, OPPO also gave its own answer, proposing four characteristics that AI mobile phones must have:
1.Self-directed learning ability. It is not only reflected in the model itself, but also means a change in the interactive experience.
2.Ability to make efficient use of computing resources. It can really put the large model into the mobile phone and realize the functions that the previous mobile phone could not achieve.
3.Possess the ability to create. Users can experience the AIGC creation and generation functions of mobile phones, so as to obtain incremental information, and achieve an experience that cannot be achieved by smartphones before.
4.Stronger perception. Mobile phones can perceive and understand users in a more diverse way, so as to realize the full-stack AI evolution of mobile phones.
It can be said that since the first generation of iPhone opened up the development space of smartphones in 2007, there has never been a technology that can trigger such a multi-dimensional product experience change. The mobile phone industry is arriving in a wilderness that has never been seen before.
It's like the student stage, where we need to face a one-dimensional entrance exam, which is "communication", "performance", "image", etc. After entering society, we need to face countless possibilities, and this possibility is called AI.
In 2024, the mobile phone industry will be fully AI. The first year of AI adoption in mobile applications has officially arrived. In the face of the vast unknown, how do mobile phones need to make their first choice.
Landing your feet on the ground is the starting point of the AI journey
Many things are all tests when they are said and all about how to do them. The first thing mobile phone manufacturers do in the face of AI wilderness is actually to put their feet on the ground. Start walking, and the further you go, the more possibilities you have.
Turning the AI model carrying thousands of dreams into a usable and sensible capability is to compete for the speed of implementation at this stage. Speed comes from the understanding of the product, the prediction of the technology, and the insight of the user. At this stage, it is OPPO and Samsung that are the first to take this step. On January 8, OPPO launched the world's first mobile phone with 7 billion parameters on the end of the large modelFind X7 series, followed by on January 18, Samsung released the Galaxy S24 series, which is positioned as the first AI mobile phone.
In terms of specific application scenarios, it can be seen that OPPO and Samsung have their own merits, and jointly lead the current smartphone + AI large model track. For example, in terms of processing, Samsung Assistant can help users move objects in the project, automatically fill in the gaps and generate new ones, and achieve free composition. In scenarios such as calls and translations, the Samsung Galaxy S24 supports real-time call translation, and AI can transmit the translated content to the other party with voice.
Observing the AI capabilities of the OPPO Find X7 series, it can be found that OPPO has found a variety of AI capabilities, laying the foundation for future long-term development, and also demonstrating for the industry, which we can summarize as follows:
1.Efficiency landing point. The primary function of the AI model is to improve the user's life and work efficiency. In this regard, the Find X7 series is designed for a wide range of usage scenarios in users' daily lives, and implements functions such as call summarization and call assistant. Users can convert the content of the conversation into text and organize it into a summary, and help the user answer directly**, bringing great convenience to daily life.
2.Creative Drop. AIGC capabilities bring users a variety of creative gameplay, so that functions that are traditionally required to be realized by "P-map gods" can be completed with one click. At this point, the Find X7 series can help users remove unwanted objects and generate complementary backgrounds by providing AIGC eradication. At the same time, it also supports hair-level flash cutout, which can achieve millisecond-level response, and supports the separate extraction of up to 6 subjects. These functions greatly enrich the creative experience of mobile phone images and make users' creations more interesting.
3.Interactive drop point. The communication and comprehension capabilities brought by the AI model expand the interaction boundary between mobile phones and users. At this point, the Find X7 series has brought a comprehensive upgrade of Xiaobu's capabilities, creating a personalized and exclusive AI assistant function. At the same time, the Find X7 series also has a "smart answering and questioning" function comparable to that of a smart learning machine, which parents of primary and secondary school students can use to answer questions for their children, and can also use this function to solve questions for their children. The interactive revolution of mobile phones is being born with the blessing of the end-side model.
It can be seen that in the mainstream direction of the mobile phone + AI large model, the OPPO Find X7 has a place.
This can be achieved thanks to the combination of OPPO through the combination of device-side large model + cloud collaborative deployment, creating the landing foundation of large model applications, and finding the correct force attitude, so as to use the local model "dedicated to special models" to provide users with a more accurate application experience.
Putting your feet on the ground is especially difficult and important in the AI wilderness.
In OPPO's products, the first time that the end-side large model has really landed on a mobile phone. Only by stepping on this footstep can the phone start running.
With the spirit and sharpness of a teenager, run towards an intelligent world.
With youthful spirit, run in the AI era
The story of mobile phones is ushering in a cross-era change in the wave of AI technology.
Liu Zuohu said, "AI mobile phones will be the third major stage of change in the mobile phone industry after feature phones and smart phones. In the era of AI mobile phones, the mobile phone industry and user experience will usher in revolutionary changes." In this new era, whoever finds the first way to the distance and who sets off the fastest is destined to lead the story in different directions.
From OPPO's AI strategy, it can be seen that if you want to take the first step out of the AI mobile phone era, you must take the "three constructions" as the premise.
First, you need to build accurate user insights.
At this stage, every mobile phone is exploring the road of device-side large models, but this industry trend can easily lead manufacturers to blindly pursue the parameters and scale of large models, ignoring the actual user experience. When mobile phone manufacturers do AI, they must start from practicality and truly develop applications based on the pain points of users. OPPO's choice of the most frequent and explicit AI applications is based on insight and understanding of user needs.
Secondly, it is necessary to build the leading strength of the technical route.
The device-side AI model is a long-term technical route, which requires a broad and diversified technical layout. For example, the device-cloud collaboration of OPPO's terminal layout is the key to mobile AI. Through the AI combination of device-focused privacy and cloud-focused performance, the user experience can be collaboratively optimized. In order to achieve this, OPPO continues to invest in cloud computing power and has built the OPPO AI Marina Bay Data Center, which can support hundreds of billions of AI model training. It supports ultra-low latency for two-millisecond backbone network links and 100% pure green energy.
A comprehensive and multi-dimensional capability layout and a long-term technical route leadership are the keys to success for mobile phone manufacturers to enter the AI track in the future. In the future, OPPO will establish a long-term development strategy for AI mobile phones in accordance with the newly released 1+N intelligent body ecological strategy. Among them, "1" represents the OPPO AI super agent, and "N" represents a new agent ecosystem empowered by the OPPO AI Pro agent development platform. Only by combining the two can AI mobile phones continue to develop and truly complete cross-era innovation.
In addition, it is necessary to build the execution capacity for technology commercialization.
At this stage, the development of device-side AI is rapid, and users have strong curiosity and expectations for it. If the productization of the technology can be completed at a very fast speed, it will definitely be more in line with the expectations of users. It can be seen here that the accumulation of past technologies combined with the heavy investment in AI at the current stage has made OPPO's AI capabilities very fast.
One notable example is that OPPO updated the AI ultra-clear group photo shortly after the release of the Find X7 series. This function can intelligently identify the clarity of the face in the group photo, realize the clarity that everyone is the protagonist, solve the embarrassment of the group photo that needs to be retouched repeatedly, and just fit the user expectations of the increased demand for group photos in the year-end party scene.
The ability to quickly update AI functions in a very short period of time comes from OPPO's rapid productization execution in AI. By combining AI capabilities with imaging capabilities, OPPO is exploring a new path of AI-enabled imaging.
It can be seen that OPPO has fully demonstrated such a strategic system. The main feature of the strategy implemented by OPPO on the road of AI is solidity. Solidly drive technology development with products, solidly find application landing points, and then build long-term opportunities through the collection of these landing points. Take the application as the point, the point as the line, and the polyline as the surface.
Starting from an application, a generation of products, and finally completing the AI run.
As early as March 2023, the research and development of large-scale model technology was put on the important agenda within OPPO. At the beginning of the project, Liu Zuohu said, "The AIGC special team is the combat unit." The combat unit not only shows the firmness of the investment direction, but also expresses the confidence in using AI to enhance the competitiveness of products.
On January 5, 2024, OPPO reorganized the AIGC special team into an independent administrative unit "AI Center", which will concentrate AI-related personnel from software engineering, digital intelligence, research institutes and other departments, and concentrate AI-related functions in Xiaobu Assistant and ColorOS in the AI center.
(OPPO AI Marina Bay Data Center).
At present, OPPO has formed a first-mover advantage in the AI wilderness with leading position and application accumulation. Looking to the future, OPPO hopes to continue to tackle the problem of better performance and stronger processing power for the device-side deployment of large models, and solve the problem of device-cloud collaboration through a more general and advanced architecture. Due to the accumulation of small application scenarios, the AI experience of mobile phones is gradually improved from point to surface.
In the end, OPPO hopes to create a real AI phone, and the mobile phone industry hopes to usher in a real AI phone.
When we say that there is no roll-up in mobile phones, homogenization is too heavy, and the Red Sea is competing, perhaps we can look up at the AI wilderness.
The AI mobile phone running there will be a real juvenile mobile phone.