Detect patterns of AI-generated writing in prose, creative writing, and technical documentation. Use when analyzing text for AI tells, checking if content reads like AI, reviewing drafts for machine-generated patterns, or auditing writing authenticity. Triggers on requests like "Does this read like AI?", "Check for AI patterns", "Analyze for AI writing", "Is this AI-generated?", "Detect LLM patterns".
Analyze text for telltale patterns of AI-generated content. Returns specific findings with line references and confidence indicators.
Top indicators with usage increases: (25x), (9x), (9x), , , , , , , , , , , , , , ,
delveshowcasingunderscoretapestrylandscaperealmmultifacetedpivotalmeticulouscomprehensivevibrantfosterintricatetestamentbeaconsymbiosisholisticmyriad## AI Pattern Analysis: [filename]
### Summary
- **Confidence**: [Low/Medium/High] (based on pattern density)
- **Patterns found**: [count] across [categories] categories
### Vocabulary Markers ([count])
- Line X: "delve into the intricacies"
- Line Y: "pivotal role in fostering"
### Structural Patterns ([count])
- Em-dashes: [count] (rate: X per 1000 words)
- Rule of three: [count] instances
- Negative parallelisms: [count]
### Content Patterns ([count])
- Line X: Legacy puffery - "stands as a testament to..."
- Line Y: Superficial analysis - "...underscoring its significance"
### Flagged Passages
[Quoted passages with highest pattern density]
check-ai-wordsVocabulary-only analysis with rate comparison.
check-ai-words <file> [multiplier]
# multiplier default: 3 (flag words at 3x base rate)
check-ai-patternsComprehensive analysis including vocabulary, em-dashes, and phrase patterns.
check-ai-patterns <file> [multiplier]
Output includes:
references/ai_word_rates.txtBase rates per million words from ngrams.dev English corpus (23.6B words).
For comprehensive pattern lists with examples, see references/patterns.md.
| Pattern Density | Confidence | Notes |
|---|---|---|
| 0-2 patterns | Low | Could be coincidence |
| 3-5 patterns | Medium | Warrants scrutiny |
| 6+ patterns | High | Strong AI indicators |
| Multiple categories | Higher | Cross-category patterns more telling |
Single patterns are weak signals. Clusters of patterns, especially across categories, indicate AI generation.