Last month, the Nubia NaviX Ultra officially went on sale. Starting at CNY 5,999, and CNY 5,499 after the national subsidy, Nubia subsequently announced that its first-day sales revenue "surpassed 100 million in one second."
This mass-produced device, dubbed the "second-generation Doubao smartphone," focuses on "getting things done for you." For example, if a user says, "Help me buy the cheapest power bank," the phone will automatically open the app, compare prices, and place the order; or if a user says, "Help me order fried chicken delivery," the phone can open the food delivery app on its own and provide the best choice based on your preferences.
However, after purchasing the phone, many consumers found that while the AI assistant can open apps like WeChat, JD.com, and Meituan, it cannot continue to perform in-app operations. This is the case for at least the next 30 days, because Doubao has introduced the Screen Automation Operation Declaration Agreement (SAEP), which allows third-party apps to declare whether they accept GUI agent operations, with a 30-day public notice period. If mainstream apps explicitly refuse after 30 days, this AI smartphone will truly become a "ByteDance-exclusive device," as netizens have joked.
From the first to the second generation, in order to seize the right to define the next generation of smartphone interaction, ByteDance may not give up its attempts in AI smartphones. But phones are ultimately used by humans. Setting aside technical issues, do users really need an AI smartphone that can get things done and improve efficiency? This is also something tech giants need to consider.
Entertainment Above All
Counting from the release of the first-generation iPhone in 2007, smartphones have been developing for nearly 20 years. Manufacturing technology has become increasingly mature, and functions have become richer. In this process, our dependence on mobile phones has also grown stronger. But what exactly are most people doing with their phones?
The data is very intuitive—entertainment. According to the "China Online Audio-Visual Development Research Report (2026)," the per capita daily usage time of online audio-visual users is 201 minutes, about 3 hours and 21 minutes. Among them, short videos have become the only segmented application maintaining "dual growth" and the application with the longest usage time.
How much do Chinese people love watching short videos? In October last year, a data report by QuestMobile showed that among the mobile internet usage habits of post-00s users, "watching videos" has become the absolute mainstream with a time share of 37.1%, spending an average of about 2.7 hours a day on mobile videos. The report also pointed out that besides "watching videos," mobile social networking (25.2%) and mobile gaming (9.7%) constitute the "top three hobbies" for post-00s smartphone usage.
In terms of user time, a typical change is that Douyin has taken the first place from WeChat. Behind this is undoubtedly a change in public demand; people no longer need more "efficient tools," but rather more "emotional outlets." This drives users to be more willing to spend their time on instant, relaxing, and low-cost pleasures.
This raises a question: when we watch short videos, play games, or chat with friends, do we need AI to help improve efficiency?
The answer is most likely no. Whether it is watching short videos, streaming shows, socializing, or listening to music, these commonly used apps on mobile phones have undergone countless iterations, allowing users to enjoy pleasure with almost foolproof operations. Especially for short videos, all it takes is continuous swiping with your fingers, omitting even the search process.
In other words, most functions of current smartphones have actually reached a level of ultimate efficiency in themselves.
Of course, besides leisure and entertainment, we also need to use mobile phones for shopping, ordering food delivery, traveling, and other activities. When shopping online, AI might make the selection process more efficient, thereby reducing time costs. But in fact, the joy of online shopping for many people lies in enjoying the process of selecting and comparing products, which also gives them a greater sense of satisfaction the moment they open the package. The same goes for food delivery.
Actually, we can think about it: among the things we often do on our phones, is there anything that cannot be completed without AI, or cannot be done better with it? Perhaps only some work-related tasks, such as writing weekly reports, editing pictures, opening PPTs, etc. But these needs are too low-frequency, and many people don't even need to use their phones for work at all.
AI taking over the phone to get things done for users essentially focuses on improving efficiency, saving time and effort. However, a large number of users who cannot live without their phones and spend excessive time on them are precisely those who lack time the least.
Leaving Everything to AI?
A couple of years ago, a term went viral on the internet—"brain rot," referring to the phenomenon of mental and intellectual decline caused by the excessive input of digital information, especially low-quality information from social media. This originated when Oxford University Press in the UK selected "brain rot" as the word of the year, sparking global discussion.
The rise of this concept means that external doubts about the fragmented content controlled by algorithms have risen to the public level, which has also triggered widespread public anxiety about the degradation of abilities. Online, we can see various contents criticizing people's overuse of mobile phones and social media. Everyone may have realized to some extent that they are spending a bit too much time in front of this screen.
Last year, there was also a paper in Psychological Bulletin studying the relationship between watching short videos, cognitive ability, and mental health. The research team compiled 71 existing studies covering 98,299 participants, and then conducted a systematic review and meta-analysis. The conclusion showed a moderate negative correlation between short video usage and cognitive function (r = -0.34), meaning that the more frequently one watches, the worse the cognitive performance tends to be. Specifically, the most severely affected cognitive abilities are attention and inhibitory control.
Watching short videos every day will affect cognitive abilities. But what about leaving things to AI, letting AI assistants complete tasks on your behalf, or turning to AI for help whenever encountering problems?
A reasonable speculation is that we might become lazier in thinking.
Some time ago, a paper titled "The Learning Penalty of Generative AI: Evidence from Secondary Education in China" tracked 26,811 middle and high school students in a county in central China. It found that after using generative AI to outsource homework, students' homework scores improved by about 18% in the short term, but their monthly exam scores dropped by an average of 20% within six months.
This paper even reported the results of the high school entrance examination and the college entrance examination. Among students who had used AI cumulatively for at least two years, high school entrance examination scores dropped by about 24%, and college entrance examination scores dropped by about 18%.
For students, efficiency is only a means; the improvement of learning ability is the ultimate goal. Using AI tools only gives them the results, while the thinking process is replaced. Although the purpose of adults using AI tools is to improve efficiency and free themselves from repetitive labor, once they become increasingly dependent on the convenience and efficiency of AI, the resulting outcome may be consistent with that of students.
Because this time, the technological revolution is triggered by generative AI, which is a tool that allows us to outsource cognition and thinking more than ever before. Just by entering a prompt, you can instantly get almost any answer, write papers or code, create art or design works. It can be said that this is a "dissolution" of the human thinking process, perfectly catering to human laziness.
This is even more true for AI smartphones. In the future, with just a single sentence, we can sit back and wait for the results. But by then, instead of being freed from meaningless labor and devoting themselves to valuable things, many people will just become lazier and lazier.
Sense of Losing Control
In the past two years, emotional consumption has boomed, and the happiness economy remains popular. Internet buzzwords such as "finding oneself in later years," "prefrontal cortex-friendly," and "life odyssey" have emerged one after another. Behind all this is an obvious and universal psychology: as the external world becomes increasingly uncertain, young people need to regain a sense of certainty for themselves.
For example, they say they have no money on one hand, but are willing to buy a cup of coffee costing tens of CNY on the other; they shout about consumption downgrade on one hand, but are willing to pay for a small toy, a fragrance, a performance, or a ritualistic meal on the other.
This psychological mapping has long been reflected in consumer behavior, and to some extent, it also represents many people's attitude towards their current lives.
The opposite of a sense of certainty is a sense of losing control. The premise of AI smartphones getting things done is to let the AI assistant replace the user and navigate across various apps. But for such a private device containing a large amount of privacy information, are users really completely reassured to let AI take over? This is one question. Another question is, if we rely on AI for a long time, we may gradually develop a sense of losing control: it is not "I" who is controlling my phone, but AI.
For example, in the future, a user says, "Help me book a hotel," and quickly gets a result saying "Booked." The user often only sees the conclusion, without seeing what the AI filtered out or why it chose this one.
This is actually an impact on past consumption habits, which can easily cause discomfort or resistance. More importantly, in the process of AI making evaluations and comparisons, which products it excluded, what criteria it used for ranking, and whether the recommended spots involve commercial cooperation—users are unaware of this information.
Users not knowing this will naturally lead to distrust.
Of course, when we watch short videos, we are also unclear about the mechanism of algorithmic recommendations. The platform controls the ranking weights and traffic rules, but this opacity can be tolerated by users because the algorithm only affects a list of "what you see." If you like it, you watch; if you don't, you swipe away. But once an AI assistant takes over the phone, accompanied by the commands is a series of "executed actions." If the user is dissatisfied with the result, they need to re-enter the command and regenerate the result.
Once this process is repeated a few times, users may begin to question or become dissatisfied with the efficiency of the AI assistant.
Moreover, watching short videos is free, while letting AI filter information is mostly for the purpose of consumption. When it involves spending money, users will inevitably be strict about the experience and efficiency.
Furthermore, on a mobile phone, an AI assistant that can get things done and do them well must necessarily understand all aspects of the user's information. Not to mention whether tech giants are willing to accept agent access, such an "existence" that knows you better than you know yourself may not represent convenience for many users, but rather fear.
We cannot deny that, in the current situation where the smartphone market is increasingly sluggish, an AI smartphone that can get things done is a positive exploration to break through the bottleneck of innovation. But in the competition among giants, for the mobile phone, which is the "second self" of humanity, any innovation should first ask whether users need it.
Dao Zong You Li, observer of technological innovation and business trends. Deeply engaged in the tech and business field for 15 years, fully crossing the three major transformation cycles of the PC internet, mobile internet, and AI industries. This article is original content. Reproduction, extraction, and modification without authorization are prohibited. Sharing and forwarding are welcome.