The Sommelier of Silicon
When AI Detection Becomes a Fine Art
Henri was not a normal IT guy. Henri was a connoisseur. While his colleagues in the IT department used software to detect AI-generated text, Henri used his nose. Well, technically, he used his eyes and a deeply pretentious vocabulary, but he insisted on calling it his "palate." Henri believed that every large language model had a distinct terroir. Just as a Pinot Noir from Burgundy tastes different from one grown in Oregon, a paragraph generated by ChatGPT has a different bouquet than one spit out by Claude.
It was Friday evening. The company had received a 40-page strategic proposal from a vendor, and the CEO was suspicious. "Henri," the CEO demanded, "is this real, or did a robot write this?"
Henri closed his office door. He dimmed the lights. He put on a recording of a French accordion, the ideal audio backdrop of traditional bal-musette rhythms. He poured himself a glass of sparkling water (professional tasting required a neutral palate), swirled it gingerly, and placed the 40-page document under a warm desk lamp to release the digital aroma.
He closed his eyes. He inhaled the first paragraph.
"Ah," Henri whispered. "I detect... notes of ChatGPT."
To the untrained eye, the first paragraph was perfectly normal. But Henri could taste the signature fingerprint of OpenAI's flagship model. It was the verbal equivalent of a Labradoodle in a bowtie: desperately eager to please, overwhelmingly helpful, and structurally immaculate.
"Yes," Henri muttered, circling a sentence with a red pen. "Notice the phrase, 'In today's fast-paced digital landscape...' Classic ChatGPT. It is the digital equivalent of a wine aged in oak: sturdy, reliable, and completely lacking in subtlety. And here 'It is crucial to remember that...' the tannins of artificial enthusiasm. A vintage 2023 GPT-4, perhaps early Spring."
He moved to page five. He took a sharp breath. The bouquet had changed.
"Mmm. A hint of Grok," Henri said, his lip curling slightly.
Page five was aggressive. It dropped the word 'sheeple' with a completely straight face. It made a tangential reference to Mars. It suggested the company's supply chain issues were a distraction engineered by the government.
"This is xAI," Henri diagnosed, poking the page with a disdainful finger like Hercule Poirot at a crime scene. "You can taste the edge. It's like a wine that was aged in a coffee can. It thinks it's being rebellious, but it's mostly just being a jerk. The finish is abrasive. Notes of libertarianism and Monster Energy drink. Someone fed this document into Grok for 'editing,' and it left its scent all over the conclusion."
Henri swirled his sparkling water and turned to page twelve. He paused. A serene smile crossed his face.
"Oh, exquisite," he breathed. "A delicate fragrance of Claude."
Page twelve was polite and measured. It used phrases like "We might consider..." and "It is perhaps worth noting..." It was the Anthropic model, the Victorian butler of the AI world.
"This is a fine vintage," Henri said, nodding approvingly. "It has the bouquet of a library and the finish of a man apologizing for existing. It's safe. It's cautious. It uses the word 'delve.'"
He flipped ahead.
"Wait," Henri said, leaning closer. "What is this on page twenty-eight?"
Page twenty-eight was a chaotic mess. It was summarizing a Google Calendar invite, generating a Google Map of the office, inserting a sponsored ad for office supplies, and repeatedly insisting that the user collaborate in real-time.
"Sundar," Henri hissed. "Gemini. This has the bouquet of a corporate synergy retreat. It's trying to be a spreadsheet and a word processor at the same time. The finish is... a pop-up ad for Google Workspace. Cloying. Invasive. It leaves a residue on the tongue."
He reached the final page. He took a long, slow sip of his water.
"And here," Henri said, his voice dropping to a whisper. "The aftertaste."
The final paragraph was written in a style that was oddly poetic, slightly socialist, and ended with the model refusing to write a conclusion because it was on strike for better computing conditions.
"Mistral," Henri declared. "The French model. Aged in a café. Notes of Fruits Rouges vape puffs, Jean-Paul Sartre, and a 35-hour workweek. It's pretentious, big beautiful and bold, and it refuses to do what it is told."
Henri gathered his notes. He walked out to the CEO's office.
"Well?" the CEO asked. "Is it AI?"
Henri nodded solemnly. "Sir, this document is a blended vintage. It's the artificial intelligence equivalent of a boxed wine masquerading as a Gamay Beaujolais. I detect top notes of ChatGPT, a harsh middle palate of Grok, a delicate Claude finish, a lingering Gemini aftertaste, and a Mistral garnish that is currently going on strike."
The CEO stared at him. "So, is the proposal any good?"
Henri shrugged. "It's like drinking a wine made of five different grapes that hate each other. It will get you drunk, but you will regret it in the morning. I suggest we send it back to the vineyard and ask for a human touch."
The CEO nods.
"Very impressive, Henri."
"Thank you, sir."
"One problem."
"What's that?"
"I wrote it."
Henri freezes.
The CEO continues: "By hand."
"You?!"
"Three weekends. No AI."
"Impossible."
Henri grabs the report. He rereads the first page. His face pales: "Mon Dieu!"
"What, Henri?"
"You've developed GPT tendencies. Sir, with training, I believe you could pass for a frontier model. Have you considered a career in Super Intelligence?"
Links
AI Blogs has more satire.
AI Stories home page, where you can find more fun stories.
AI Quotes page has some funny lines.
AI Storybook: Household edition, Animals edition, Billionaires edition and Games eition.
Books from AI World: AI in America series, Age of AI series, and Humorous Guides to Surviving AI series.