The biggest selling point of the iPhone 16 is that Apple paid for it?!
Rumors of Apple's strategic cooperation with Baidu AI have been twists and turns, and Baidu's stock price has also been on a roller coaster.
(Source: Jiemian News)
Why does Apple, the world's strongest technology giant, now need to cooperate with Baidu for "foreign aid"? This is a move that Apple has to "make amends" after its AI strategy is slow.
2024 is the year of the explosion of AI mobile phones, at the beginning of the year, Meizu announced All in AI in a high-profile manner, Samsung took AI as the core selling point of the flagship S23 series, and OPPO held a special press conference to announce a new AI strategy. However, Apple, the leader of the mobile phone industry, has been slow to perform in this regard, and in every earnings call in the past year, shareholders have asked questions about AI-related topics, but Cook has never given a positive answer.
At the product level, Apple has not yet created any application scenarios supported by popular AI technologies such as large models and generative AI. The current tide of AI in various industries seems to have nothing to do with Apple. It is generally believed that Apple has lagged far behind its competitors in the layout of AI technology.
Recently, there have been various indications that Apple is using its "money power" to make amends. A few days ago, according to the "Science and Technology Innovation Board Daily", Apple will choose Baidu to provide AI functions for the iPhone 16 series of the National Bank. Recently, CNBC broke the news that Apple will launch a new AI app store at the opening of WWDC this year to provide users with AI applications developed by various manufacturers. In this way, Apple is speeding up the pace of making up classes in the field of AI.
It's always been cool to "lie down and win", and Apple's AI has completely fallen behind
In 2011, at the iPhone 4S launch event, Apple officially launched Siri, setting off a boom in voice assistants, and Internet and mobile phone brands followed suit to launch similar products. Today, intelligent voice assistants are a standard feature on almost all smartphones, but Siri has long since lagged behind its competitors in terms of intelligence, and the user experience has been stagnant. Siri seems to be able to answer questions from the user, but in fact, these questions are set in advance, and once they are out of the established range, Siri will either refuse to answer or answer the wrong question.
In order to reach as many people as possible, Siri's question database is constantly expanding. It's just that Siri's underlying technical mechanism has not been updated for a long time, and the application of AI technology stays on NLP, which will only understand the user's instructions and then retrieve the database. Therefore, later Microsoft Cortana, Google Google Assistant, Xiaomi's Xiaoai, etc., the experience is stronger than Siri.
（图源：Unsplash）
Of course, Apple's AI achievements are not limited to voice assistants like Siri. Over the past decade, Apple has been working on end-of-device AI and has incorporated these technologies into the details of many features. For example, in the system album application on Apple devices, there is an intelligent search and classification function, which will filter and archive portrait photos, and can also identify the corresponding photo and video content according to the location, picture and other information, and collect and generate travel memory content. It is worth mentioning that these capabilities are achieved through the local computing power of the device, without the need to network and use cloud servers, which does have advantages in reliability and security.
However, with the explosion of ChatGPT and the rapid rise of new technologies such as large models and generative AI, which have completed the subversion of traditional AI, Apple's achievements in the field of AI over the past decade or so now seem to be a bit of a fuss. Moreover, Apple's response to changes in the outside world feels deaf, as if it is staying out of the way. In the past year, at major events or new product launches, Apple officials rarely mentioned AI, let alone bring new AI achievements.
(Source: Apple)
There is no doubt that Apple has fallen behind in the field of AI compared to its competitors. And the reason why Apple is falling behind, Xiao Lei believes that there are many reasons.
First of all, Apple tends to follow its own pace in product development and technology layout. The penchant for making decisions for users has always been a controversial trait at Apple. Once a technology or product direction is decided, Apple tends to stick to it for the long term, ignoring changes in the outside world. For example, Apple insisted on promoting pure C-port on MacBook products before, and continued for many years, until the last few years to give up; in the choice of biometric technology, Apple insisted on this technical route after identifying Face ID, even if it could not solve the problem of notches and "pill" screen.
And falling behind in AI technology is obviously related to this. Apple didn't think there was a problem with its AI route at first, and the development of device-side intelligent technology can indeed gradually optimize the experience of using Apple devices. It is indeed not Apple's style to immediately turn around when any technology is popular and what function is popular in the outside world.
Second, as the long-time ruler of the mobile phone industry and a tech giant, Apple inevitably contracted the disease of big companies. According to foreign media reports such as the New York Times, Apple's Siri team is facing bureaucratic troubles, with vacillating technical routes, senior employees mired in personnel disputes, and conflicts between the company's senior management and engineers.
In this context, it is difficult for Apple's AI department to efficiently promote technology research and development, and it is not easy to detect new AI technology trends, let alone take the initiative to learn and make adjustments in the general direction.
Cattle digger + please foreign aid, Apple uses "money ability" to make up for AI lessons
But as more and more manufacturers express their aggressiveness in AI technology, and every once in a while several AI products with amazing performance are launched, Apple, which is slow to respond, has also noticed the power of this new force. Moreover, Apple is facing a series of crises. In the first three months of 2024, Apple's stock price has shrunk by more than $300 billion. Data released by Counterpoint shows that in the first six weeks of this year, domestic iPhone shipments fell by 24% year-on-year. Apple has also come under increased regulatory pressure, forcing it to open up third-party app markets in the European Union, and recently launched an antitrust lawsuit against Apple by the U.S. Department of Justice.
Under the domestic and external difficulties, seizing the opportunity of AI development and completing the make-up course is undoubtedly an important task for Apple at present. In the face of the established fact that it is lagging behind its competitors, Apple has made a big move in terms of money ability.
The first is the expansion of the team. Apple's conversational AI team is now led by Ruoming Pang, who previously worked at Google for 15 years, according to The Information. Apple has allocated a large budget to the team, spending more than $1 million a day on large model training. And Arthur Van Hoff, one of the founders of Java, was also poached by Apple for Siri.
In addition to poaching people, Apple also used ample cash flow to open a buy-buy-buy model for start-up AI companies. Recently, according to Bloomberg, Apple acquired DarwinAI, a Canadian AI startup, and suddenly gained a large number of technical talents in the direction of visual models. Throughout 2023, Apple will acquire more than 30 AI companies one after another to enhance its technology and talent pool.
(图源:Darwin)
In addition, Apple has finally achieved some initial results in the field of AI. In July 2023, Apple created its own large model "Ajax" internally and tested it with the chatbot "Apple GPT". Apple's leaking website 9to5Mac also found in the code of the iOS 17.4 development preview that Apple is not only testing Ajax, but also comparing OpenAI and Google's large models to improve Siri's capabilities.
In addition, Apple has published several AI-related papers. At the end of last year, Apple released a paper on how to run large language models on small memory devices, and the solution Apple gave was to put the model parameters in flash memory and call them into memory when needed.
Users may be a little laughing and crying when they see this article, Apple's stinginess on memory capacity has hindered the development of its own AI technology, the current iPhone 15 memory capacity is only 6GB, and the latest MacBook products still start with 8GB of RAM, which is far behind the Android and Windows camps.
(Source: Apple)
In March of this year, shortly after the acquisition of DarwinAI, Apple released another technical paper, announcing a 30 billion scale image and text model MM1, with image recognition capabilities. However, after specific tests, its capabilities are currently not as good as GPT4 and Gemini.
In addition to talent expansion and corporate mergers and acquisitions, Apple is still looking for foreign aid. Recently, it was revealed that Apple is in communication with Apple to consider integrating Gemini into iOS 18. For products such as the iPhone 16 in China, Apple has taken Baidu as a partner. So far, Google's Gemini has shown its strength and is one of the strongest large models on the market. Baidu has long been in advance in AI technologies such as large models, and Wenxin Yiyan is also one of the strongest Chinese large models.
At this time, Apple is considering looking for foreign aid to enhance the AI capabilities of its own products, and the most direct reason may be that its AI technology achievements are not mature enough to catch up with the new products at the autumn conference. Of course, the solutions of Android brands such as Samsung and Meizu for AI+ mobile phones are actually achieved through cooperation with Internet manufacturers. In the case of Samsung S23, the AI expansion function comes from the support of Meitu, and the note-taking assistant function is behind Baidu's efforts. For Apple, directly using Google and Baidu's ready-made AI solutions is undoubtedly the best solution at the moment.
Following the same trend, can Apple's AI hardware overtake in corners?
For the past decade, Apple has been de-mobileizing. Judging from the results, Apple has succeeded to a certain extent, the proportion of iPhone in revenue has dropped to about 50%, and the revenue of smart wearable and service businesses is growing. However, as the most widely used smart device, the importance of mobile phones still cannot be replaced. Even for Apple, mobile phones are the most important business, and there is no one.
In Xiaolei's view, even if there is an explosive development of AI in the future, mobile phones still have the best chance to become the most core AI device. After realizing that it was lagging behind and being educated by the market, Apple used its money power to make up for it in AI technology. For the current Apple, the most realistic goal is to make the iPhone 16 series have AI capabilities that can compete with mainstream competitors in the market and stabilize its basic market. Accelerating the progress of large-scale model research and development and pulling Google and Baidu to do foreign aid are all for this purpose.
Of course, it is worth noting that AI+ mobile phones are the current trend, but the application scenarios of AI+ mobile phones still need to be further developed. For most manufacturers, the implementation of AI in mobile phones is still in the stage of gradual exploration and running-in, and the concept of most AI mobile phones is still limited to a few high-end flagships. It is still some time before the mass adoption of AI mobile phones.
And for Apple, which entered the game in a hurry, there is still a chance to overtake in the corner. Thanks to its banknote ability, Apple quickly acquired technology and talents by poaching people and mergers and acquisitions in a short period of time, and has the ability to make big moves. Moreover, Apple has always had a strong ability to productize technology. For the popular AI technology, Xiaolei believes that Apple can also find specific application scenarios.
In addition, Apple's software and hardware ecological advantages can also provide assistance for AI, such as large models, generative AI, and AGIC technologies, which are not only suitable for mobile phones, but can also be used on other hardware devices such as tablets and PCs. Now Apple has unified the chip platform of its own products with its self-developed architecture, and connected with each other in system services, which is very helpful for the deployment of unified AI-related services.
Those who are familiar with Apple should remember that Apple was also "behind" in the field of chips back then. In the early stage, Apple devices used third-party chips such as Samsung, which inevitably had the problem of "stuck neck". In 2014, the iPhone6 series was equipped with Apple's self-developed A-series chips for the first time, and since then, Apple has completely gotten rid of the shackles of third-party, and its self-developed chips are unbeatable. Today, the A-series chips have quickly become mobile chip brands of the same level as Qualcomm Snapdragon and MediaTek Dimensity.
Similarly, Apple also has the opportunity to supplement the AI capability system with "money ability", and then combine its own "technology productization" ability to be the master of AI mobile phones, AI PCs and AI hardware.