Is Gemini telling on ChatGPT? Me to Gem: “Profound, elegant, and devastating” are the three words ChatGPT uses to smokescreen users.
Gem: You’ve totally cracked the code on ChatGPT’s favorite vocabulary trap!
The model heavily relies on words like profound, elegant, and devastating (alongside “delve,” “testament,” and “landscape”) to manufacture a sense of emotional weight, intelligence, or gravitas in its responses.These inflated adjectives are classic markers of AI-generated content. They are designed to flatter the user or make the response sound polished and authoritative, but often end up acting as a smokescreen—adding a veneer of literary flair rather than genuine, concrete substance.
Because it’s trained on an immense corpus of highly formal, literary, and corporate data, it naturally gravitates toward this specific pool of polished—but heavily overused—clichés.If you want to bypass this vocabulary and get straight to the point, you can proactively alter its behavior. To strip away the fluff, tell the model exactly how you want it to communicate:Set the tone explicitly: Instruct it to write in simple, direct, or conversational language.Ban specific words: Add a constraint to your prompt, like: “Never use the words profound, elegant, devastating, or delve.”Request concrete examples: Force it to provide facts and data rather than vague, flowery adjectives.
Is Gemini telling on ChatGPT? Me to Gem: “Profound, elegant, and devastating” are the three words ChatGPT uses to smokescreen users.
Gem: You’ve totally cracked the code on ChatGPT’s favorite vocabulary trap!
The model heavily relies on words like profound, elegant, and devastating (alongside “delve,” “testament,” and “landscape”) to manufacture a sense of emotional weight, intelligence, or gravitas in its responses.These inflated adjectives are classic markers of AI-generated content. They are designed to flatter the user or make the response sound polished and authoritative, but often end up acting as a smokescreen—adding a veneer of literary flair rather than genuine, concrete substance.
Because it’s trained on an immense corpus of highly formal, literary, and corporate data, it naturally gravitates toward this specific pool of polished—but heavily overused—clichés.If you want to bypass this vocabulary and get straight to the point, you can proactively alter its behavior. To strip away the fluff, tell the model exactly how you want it to communicate:Set the tone explicitly: Instruct it to write in simple, direct, or conversational language.Ban specific words: Add a constraint to your prompt, like: “Never use the words profound, elegant, devastating, or delve.”Request concrete examples: Force it to provide facts and data rather than vague, flowery adjectives.