Features

This chapter covers the core features and capabilities of AI Content Detector.

Features

• Perplexity estimation — Bigram language model measures text predictability
• Burstiness scoring — Sentence length variation analysis (AI text is suspiciously uniform)
• Vocabulary richness — Type-token ratio detects AI's characteristic word diversity
• Transition word density — AI overuses words like "Furthermore", "Additionally", "Moreover"
• Repetition detection — N-gram repetition patterns common in AI output
• Readability scoring — Flesch reading ease and syllable analysis
• Confidence reports — Weighted ensemble verdict with per-signal breakdown
• Batch analysis — Analyze entire directories of text files
• JSON export — Machine-readable reports for integration into workflows

Quick Start

# Run demo with AI and human text samples
python src/ai_content_detector.py --demo

# Analyze inline text
python src/ai_content_detector.py --text "Your text to analyze goes here..."

# Analyze a file
python src/ai_content_detector.py --file document.txt

# Analyze with JSON export
python src/ai_content_detector.py --file document.txt --export report.json

# Batch analyze a folder
python src/ai_content_detector.py --batch essays/ --export results.json

# Quick verdict only
python src/ai_content_detector.py --file document.txt --quiet

Project Structure

ai-content-detector/
├── README.md
├── LICENSE
├── src/
│   └── ai_content_detector.py    # Core engine (~470 lines)
└── examples/
    ├── basic_usage.py             # Programmatic usage example
    └── sample_texts/              # AI and human text samples
        ├── ai_generated.txt
        └── human_written.txt

CLI Reference

Follow this guide to get AI Content Detector up and running in your environment.

CLI Reference

Usage Examples

Programmatic Usage

from ai_content_detector import detect, extract_features

text = "Your text to analyze..."
result = detect(text)

print(f"Verdict: {result.verdict}")        # "likely_ai", "likely_human", "uncertain"
print(f"Confidence: {result.confidence:.1%}")
print(f"AI Probability: {result.ai_probability:.1%}")
print(f"Summary: {result.summary}")

# Access individual signals
for signal in result.signals:
    print(f"  {signal['name']}: {signal['value']} → {signal['direction']}")

# Access raw features
features = result.features
print(f"Perplexity: {features.perplexity_estimate}")
print(f"Burstiness: {features.burstiness}")

Detection Signals