Is ChatGPT Changing How We Talk And Write

The ChatGPT Effect on Everyday Language

Recent studies have highlighted a fascinating and somewhat controversial trend: words and stylistic choices favored by Large Language Models (LLMs) like ChatGPT are increasingly appearing in everyday human speech and writing. Researchers point to a noticeable uptick in the use of words such as “delve,” “intricate,” “surpass,” “meticulous,” and “garner.” This phenomenon suggests that as we interact more with AI, we are subconsciously adopting its linguistic habits. This has sparked a widespread conversation about the evolving nature of communication and the blurring lines between human and machine-generated text. Are we witnessing a subtle linguistic shift, or is this a sign of something more profound about our relationship with technology?

The Great Em-Dash Debate

One of the most heated debates revolves around a single piece of punctuation: the em-dash (—). Many argue that the frequent and proper use of the em-dash is a dead giveaway for AI-generated text. The theory is that while AI consistently uses the formal, unspaced em-dash, most humans opt for a simpler, spaced hyphen (-) or a double-hyphen (--), largely because they don't know the keyboard shortcut for the real thing.

However, this assumption has been met with strong resistance. Numerous people, particularly long-time users of Apple products, pointed out that keyboard shortcuts for em-dashes (like Option+Shift+Hyphen on a Mac) and auto-correct features in software like Microsoft Word have made them easily accessible for decades. For many writers, editors, and typographically-minded individuals, the em-dash has always been a staple. This has led to a peculiar form of paranoia, where people who have used proper punctuation for years are now actively avoiding it to prevent their writing from being dismissed as “AI slop.”

The Forbidden Vocabulary List

Beyond punctuation, a new set of “forbidden words” seems to be emerging. Words like “therefore,” once a standard for formal communication, are now being flagged by some as AI tells. One user shared that their company now has official guidelines to avoid such words, fearing they make communications sound robotic. This creates a challenging dilemma: do we simplify our language and risk being less precise, or do we use a richer vocabulary and risk being mistaken for a bot?

This has been particularly frustrating for those who have always had a more formal or academic writing style. One commenter aptly noted, “As someone who writes above a fifth grade reading level, this whole thing has been so depressing. It's like Idiocracy-level. People are going to assume I'm using AI because I use the word ‘intricate’?” The fear of misidentification is pushing some to consciously “dumb down” their language.

Proving Your Humanity in a Digital World

This growing suspicion has led to an ironic and slightly dystopian conversation about how to prove one's humanity online. Some have suggested intentionally inserting spelling mistakes or grammatical errors into their writing to signal authenticity. The logic is that since LLMs are designed to be grammatically perfect, a small error serves as a human fingerprint. Others are considering avoiding common AI patterns, like using lists of three examples, in favor of two or four.

This push for deliberate imperfection highlights the absurdity of the situation. As one person put it, “What a time to be alive in which a small spelling or grammar mistake is a good sign of authenticity.” We are entering an era where being a clear, articulate, and grammatically correct writer may paradoxically work against you in certain online contexts.

Are We Blurring the Lines Ourselves?

Ultimately, this linguistic bleed-over may be an inevitable consequence of our exposure to AI. Humans are naturally imitative creatures; we adopt the slang, accents, and communication patterns of those we interact with. With millions of people now using LLMs daily for everything from drafting emails to exploring creative ideas, it's no surprise that the AI’s style is rubbing off on us.

It’s also worth remembering that LLMs learned their style from us in the first place. Their penchant for formal language and proper punctuation comes from being trained on a massive corpus of high-quality human writing, including books, academic articles, and well-moderated online discussions. The question remains whether this feedback loop will lead to a richer, more literate society or a homogenized “model collapse” where human and AI writing become indistinguishably bland. As one person noted in a discussion about this phenomenon, “I see no reason why someone wouldn't want to copy their style or vocabulary if they want to improve their communication skills.” Perhaps the AI isn't just changing our language—it's holding up a mirror to the most articulate parts of its training data.