"This is not just art, but the science of application."
"This is not just software, but a technological revolution."
"This is not just an update, but a revolution of confidence."
The previous phrases are sentence structures currently widespread in many writings onthe internet.At the same time, Barron's, a magazine affiliated with The Wall Street Journal, reported that during 2024-2025, the phrase 'not just ... but ...' increased abnormally in documents from many major companies in the United States.
Besides such phrases, AI's tone can be noticed in many aspects, such as sentences with fixed patterns that sound grandiose, as if reading a statement from a prominent figure, yet feeling hollow and full of some hard-to-explain oddities, perceived instinctively.
These reflect that people increasingly rely on AI-generated writing today, which often annoys readers and, in the long term, may narrow human language to repetitive sentence forms or a few repeated words, making it difficult for people to express ideas deeply and diversely.
Meanwhile, companies or organizations risk losing credibility. Therefore, we must understand AI's language and tone to cope with it and define appropriate boundaries and roles for AI use, so human tone remains, preserving the value and emotion in the text we compose.
What is AI language like and why is it constructed this way?
Creative artificial intelligence, or Generative AI, is a language model (LLM) that works by collecting all words from online databases worldwide, processing them to calculate averages, then predicting the probability of the next word to explain something. This process involves reducing details to produce the most neutral words possible, making it user-friendly.
As a result, contradictions in a word or humor embedded in a sentence may be omitted, and data flattened to produce the most concise formulaic result.
Therefore, we often see repeated words like 'Delve' appearing frequently—words increasingly used in articles since AI's emergence. Similarly, words like 'Unlock,' 'Empower,' 'Elevate,' or 'Crucial' are also hallmarks of AI language.
Regarding vocabulary, AI often uses clichéd grand openings that make texts feel overused, such as 'In the rapidly evolving digital world' or 'Amidst changing dynamics.'
Besides the 'not just XXX but YYY' structure, AI often provides examples or explanations in three points, even when the third point seems unnecessary or too broad. For example, 'Office work (1) improves communication efficiency (2) facilitates learning for new employees (3) strengthens organizational culture aligned with future vision,' where the third point might be unnecessary but AI still includes it.
Most noticeably, especially in Thai, AI excessively inserts em dashes (—) mid-sentence to try to add drama or tension and evoke emotion, but misusing them and lacking context makes this irritating to readers.
AI tries to appear clever or witty by using internally contradictory sentences to attract attention, such as 'But the truth is...'—these attempt to show rebellious argumentation but often feel forced and unnatural.
However, although AI likes to present internal contradictions, fundamentally, it never produces truly harsh or deeply conflicting writing because it is programmed to be neutral and conciliatory. Another noticeable trait is it presents both pros and cons, which is beneficial if we use AI as a tool.
Concerns arise when we allow AI to produce entire texts and those words flood online databases, because thenAI trainingin future models will rely on AI-generated text to train AI again, causing our language to lose flexibility, reduced to a few average words, and future generations may lose the ability to explain deeply.
Lack of confidence in writing
Reading this, many may fear that their carefully chosen words might seem AI-generated or even avoid favorite phrases to not resemble AI language. But why should we change ourselves just because we fear similarity to machines? Does that mean humans have already lost to AI?
Actually, the words AI uses—those grandiose or overly formal ones we considerclichés—are precisely the words and phrases humans use most frequently. AI processes these and creates averages to answer us, but lacking true thought, it sometimes misuses words contextually.
Therefore, what humans should do now is not to abandon AI entirely, because the 'human truth' can still exist even as AI threatens to replace many things. What we can do is use AI properly—as a tool to shorten work time, not replace ourselves entirely, which would accelerate AI replacing humans even faster.
We, the writers here, do not reject using AI; we use it as a tool while trying to understand things. Today, new technology helps shorten time, but workloads often multiply. Despite occasional inconsistencies, humans cannot go back to the pre-AI era, just as we cannot go back to the days before computers. So now people must learn to live with AI, use it as co-creation, and ask how to use it while preserving humanity.
References: