LlmDetectionResult

Result of LLM-likeness detection for a comment. Combines word frequency analysis (deterministic) with structural pattern recognition (AI-powered).

Python

from respectify.schemas import LlmDetectionResult

PHP

use Respectify\Schemas\LlmDetectionResult;

JSON

Types are returned as JSON objects in API responses.

Fields

Field	Type	Description
llm_likelihood	float (0.0–1.0)	Probability the text was generated by an LLM. 0.0 = definitely human, 1.0 = definitely AI.
confidence	float (0.0–1.0)	How sure the system is about the verdict. Low for short comments.
signals_detected	LlmDetectionSignal[]	Specific patterns that contributed to the score.
reasoning	string	Plain-language explanation of the assessment.

LlmDetectionSignal

Field	Type	Description
signal_type	string	Category: word_frequency, structural, stylistic, or epistemic
description	string	What was detected
quoted_text	string	The text that triggered this signal (if applicable)

Example Response

{
  "llm_likelihood": 0.02,
  "confidence": 0.95,
  "signals_detected": [],
  "reasoning": "Casual tone, specific personal experience, slang — classic human comment."
}

A high-scoring example:

{
  "llm_likelihood": 0.96,
  "confidence": 0.90,
  "signals_detected": [
    {"signal_type": "word_frequency", "description": "AI-overused word: \"delves\"", "quoted_text": "delves"},
    {"signal_type": "word_frequency", "description": "AI-overused word: \"tapestry\"", "quoted_text": "tapestry"},
    {"signal_type": "structural", "description": "View from nowhere — no personal voice", "quoted_text": ""}
  ],
  "reasoning": "Dense LLM-typical vocabulary, zero personal voice, reads like generated content."
}