Since the launch of ChatGPT in November 2022, nearly 10% of the global population has become its users. AI can help people write texts, draft titles, and polish resumes, which has indeed improved efficiency in many aspects. However, linguists warn that this convenience is quietly leading human language towards "flattening". The Spanish newspaper "El Pais" recently published an article stating that humanity is living in a "ChatGPT" world. Research has confirmed that human language is being trained to be more and more like a machine, with clean and concise sentences, but lacking the clumsiness and sincerity unique to humans. The things that make humans appear 'human like' are being gradually erased by algorithms. The communication methods brought by language models such as ChatGPT, Gemini, and Claude, which are unified by algorithms, are gradually deepening this trend. Some researchers suspect that the rapid popularity of ChatGPT is making human language "too correct": short sentence structures, weak emotions, lack of sincerity, and soulless. At academic conferences where English is the lingua franca, the verb 'delve' has become a prominent 'neon light' and a symbol of people who repeatedly use AI output content. According to Ezequiel Lopez, a researcher at the Max Planck Institute in Germany, a study conducted in September last year on 280000 academic videos found that 18 months after the release of ChatGPT, the frequency of using "delve" in academic speeches, conference presentations, and 10000 AI edited papers surged by 51%. Similar words include 'complex', 'commendable', and 'meticulous'. These adjectives, which were originally rare in English corpora, now frequently appear in emails, reports, and speeches. Linguist Tom Juzek from Florida State University believes that this is a bias caused by algorithm training data. The employees responsible for reviewing language model outputs in AI companies are mostly from Nigeria, Kenya, and other places, where these vocabulary words are used much more frequently than in European and American English. So, AI unintentionally globalized regional expressions. The writing style also exhibits a high degree of consistency. Philip Sergiente, a professor of applied linguistics at the Open University in the UK, concluded that ChatGPT writing has a unique rhythm, appropriate but dull, smooth yet hollow. He found that AI prefers to insert explanatory dashes in sentences, prefers "three-point parallelism", and maintains politeness and restraint at the end of sentences. This style quickly penetrated into people's language, becoming synonymous with "rationality" and "professionalism", but it also made expression more and more like a robot. Linguist Lara Alonso from Complutense University in Madrid, Spain, pointed out that "these contents without spelling mistakes or soul make people lose the desire to reply because you cannot feel the 'sense of being alive'." While AI outsources language to become monotonous, thinking is also converging. AI generates text based on the commonality of big data, naturally pursuing a middle ground and a sense of security, so people are expressing less and less uncertain, hesitant, and rebellious views. The edges of language are polished, and the sharpness of thought disappears. Experts consider this phenomenon as the "language hypnosis effect" of ChatGPT. Regardless of whether the answer is right or wrong, AI can respond confidently, while humans are born with confirmation bias and tend to trust content that meets their expectations. As long as AI sentences are fluent, people tend to believe them. Over time, people also began to imitate this expression. In June of this year, a groundbreaking study called "The Brain on ChatGPT" at the Massachusetts Institute of Technology in the United States attracted attention. For the first time, a study has found through direct observation of brain activity that humans become lazier, less critical thinking, and more receptive to "mediated" expressions with the assistance of AI. It completely outsources thinking and makes us all more like each other, "the research author wrote. Experiments have shown that when using AI to write emails or short articles, brain activity significantly decreases, especially in the alpha wave connections related to creativity and the theta wave connections related to working memory. Participants have almost no sense of involvement as authors, and even find it difficult to recall the content they have written. Surprisingly, when asked 'what truly makes you happy,' the AI team almost exclusively answered that it is related to career success. The research author Natalia Cosmina described this as the phenomenon of 'everything being equal'. This' equalization technique 'causes language to lose its personality and emotions. Despite this, most people are still unwilling to give up the convenience brought by ChatGPT and are accustomed to the "ready to use" writing and reading mode. Linguist Adam Alexich points out in his new book "Computational French: How Social Media is Changing the Future of Language" that most people are not aware of the impact of chatbots on their language and cannot distinguish whether it is content created by AI. He believes that as human language becomes increasingly difficult to distinguish from AI language, and large model training continues to come from human texts influenced by AI, mechanized and monotonous language will become more common. A writing experiment at Swinburne University of Technology in Australia requires volunteers to write advertising copy. They wrote an average of 33 words for the first time, and after watching the version generated by ChatGPT, the average word count increased to 87 when they wrote again, with longer and more circuitous sentence structures and repetitive content. Researcher Jeremy Ruan said, 'We didn't ask them to imitate AI, but they naturally wrote it that way.'. ”For millions of people, ChatGPT has become a daily tool. People are accustomed to this lengthy, repetitive yet almost perfect expression, and consider it a "normal" writing standard. And when 'standards' are redefined, the real danger is not the loss of originality, but rather our gradual forgetting of what is unique and warm language. In this wave of "ChatGPT", how to preserve the unique imprint of human beings between efficiency and sincerity is a topic worth exploring in depth. (New Society)