Why use this
- Fast client-side triage for drafts, submissions, and moderation.
- Transparent reasons. You see the metrics that drive the verdict.
- No uploads. Works offline. Heuristic only, not proof.
Limitations and responsibility
Caviat: This tool is probabilistic. It can be wrong. It does not identify authors and should not be the sole basis for punitive decisions. Always use human review.
Input
Verdict
Metrics
Interpretive Summary
Version A:
Displays irregular rhythm and authentic linguistic noise…
🟢 Verdict: Human-like writing pattern detected.
Version B:
Shows structured, repetitive, highly polished scaffolding…
🔴 Verdict: AI-generated or AI-assisted writing pattern detected.
These findings derive from sentence variance, lexical diversity, and phrase-pattern analysis.
Preconcepts (editable heuristics)
These guide the detector. Edit and save. Toggle the checkbox above to use your saved version.
What these do
- transition_words
- hedges
- ai_boost_phrases
- human_markers
- weights
- thresholds
⚠️ AI detection is probabilistic. Treat results as indicators, not evidence. Use human judgment for high-stakes decisions.