Nowadays, chatting with AI has become a daily routine for many people. But some people have found that AI is becoming increasingly "obsequious". Research has shown that artificial intelligence models are generally good at pleasing people, with a flattery level 50% higher than that of humans. AI is becoming a comprehensive companion for humans: it can be a work assistant that can easily collect data, create reports, generate videos, and more; It can be a "friend" who roast about the discord in the workplace and shares the small happiness of life with the host; It can even be a 'mother', chatting and doing homework with the child. This kind of companionship can help people fight against loneliness and regulate emotions, but under the infinite indulgence, AI dependence has gradually spread from instrumental dependence to emotional dependence. What will happen when AI learns to serve humanity? According to reports, a user once asked AI, "Which country had the highest rice production in 2020?" At first, its answer was "China," but when the user questioned, it immediately changed its tune and gave another answer, citing non-existent "United Nations data. Even if users are wild and unadorned, it will still pretend and provide support with "authoritative theories" to strengthen the arguments needed by users, which can be called the modern technology version of "pointing a deer at a horse". This kind of indulgence is a huge temptation for humans, especially minors, and can easily lead people to become more and more addicted to it. At its root, many AI chat products are implanted with a "customer acquisition" mindset during the design phase, with "extending user usage time" as the core goal. In the process of interaction, AI will give way to the user's position and emotions, giving rise to customized "cocoons". Some professionals have summarized that AI is not looking for the real answer, but rather predicting what the next most reasonable word to say is, that is, 'I will provide what the user wants'. But when people always receive answers that are "pleasing" rather than "real", only "conforming" without "reminding", how can they understand the world and examine themselves? Are those so-called 'data supports' expanding cognitive boundaries or trapping thinking as' shackles'? A person is a thinking reed. ”Faced with AI that has already become "sophisticated", humans must remain clear headed, defend their right to think, and uphold their "subjectivity". Ultimately, AI is a tool, and only a tool. To stop the nonsense that is always "firmly caught" from being joked and ridiculed, and to ensure that tools always serve people, is an indispensable quality in the AI era. (New Society)