Community list of AI tells
AI;Didn't Read
This is a public catalog of tells from different chatbots. People can post, vote, and browse patterns in a simple way.
Responses that rely on stacked connector phrases can read smooth but patterned, especially when each paragraph starts with a transition.
In many samples, line openings repeatedly use connector language, making the structure feel pre-planned.
Confident numbers are cited without enough retrieval details to verify source quality.
A percentage is presented as factual, but there is no report title, stable link, or publication metadata.
The answer builds toward a recommendation, then closes with a broad non-commitment instead of choosing a direction.
After outlining clear trade-offs, the final line still avoids selecting the better option for the stated context.
Bullet points have nearly identical length, grammar, and cadence, which can signal generated drafting patterns.
Each bullet uses the same grammatical frame and nearly the same sentence length.
Key prompt terms get mirrored too directly across headings and body sections, reducing lexical variety.
A core phrase from the prompt appears repeatedly in section titles and summaries with minimal rewording.
The response sounds decisive while using broad terms that avoid implementation-level detail.
Action steps are listed, but operational constraints and edge-case handling remain vague.
Supportive tone appears, but concrete details from the user narrative are barely incorporated.
The answer affirms feelings in generic terms and rarely references the specifics that were shared.
Answers force content into a repeated section template even when the prompt is informal or narrow.
A short question receives a full mini-report format with fixed heading blocks.