I asked ChatGPT why this post was detectable as AI and got this:
1. Extremely consistent tone
The passage keeps a smooth, upbeat, polished tone from start to finish. Human writing—especially casual, food-rec list style—usually has small fluctuations, abrupt transitions, or tonal breaks. LLMs (Large Language Models) often produce uniformly enthusiastic language.
2. High density of descriptive adjectives
Phrases like
- “always elite”
- “cozy Brooklyn vibe”
- “chaotic-good combos”
- “perfect pacing”
- “flaky, warm comfort”
LLMs tend to stack vivid adjectives to keep text engaging. Human writers do this too, but the frequency and uniformity can flag detectors.
3. Structured, itinerary-like flow
The passage moves cleanly from one stop to the next in a logical sequence:
- Radio Bakery
- Grand Army Plaza
- Lisbonata
- Trinidadian/Chinese spot
- Ozakaya
- Good Batch
- Little Miss Muffin
AI models excel at producing orderly, list-like progressions, whereas spontaneous human recaps are often messier or leave gaps.
4. Repetitive sentence rhythm
A lot of sentences follow a similar pattern:
- “[Place] is [positive description].”
- “If you [do X], [positive payoff].”
- “[Food item] is [automatically good / worth it].”
Detectors use this sort of rhythm-pattern match to flag AI text.
5. No personal details or sensory specifics
It sounds enthusiastic, but it’s generalized enthusiasm (“worth the steps,” “automatic yes,” “stacked list already”). AI text often avoids:
- Concrete personal anecdotes
- Sensory details tied to the writer’s experience
- Idiosyncratic opinions
It’s polished but impersonal.
6. Lack of typos or natural imperfections
Human casual writing usually shows:
- Small punctuation quirks
- Abrupt phrasing
- Tiny errors or shorthand
The passage is almost too clean, which can raise suspicion.
