How it works
Paste your text
Provide the text you want to analyze
The AI detector analyzes
Our detector analyzes your text using stylometric profiling
Probability & analysis
You receive a probability score indicating whether the text was written by AI, along with a detailed analysis
AI Detector
What our detector analyzes
Linguistic analysis and statistical patterns
The detector performs an in-depth linguistic analysis including lexical diversity, syntactic complexity, and n-gram frequency analysis. It applies advanced statistical models to identify subtle language usage patterns that can distinguish between human-written and AI-generated content.
Stylometric and stylistic profiling
By applying stylometric techniques and machine learning algorithms, the detector creates a stylistic profile of the text. It analyzes factors such as sentence length, word choice, function word usage, and rhetorical devices to distinguish characteristic features of human versus AI-generated writing.
Semantic and pragmatic analysis
The detector uses advanced Natural Language Processing techniques to conduct deep semantic analysis. This includes latent semantic analysis (LSA) and transformer-based models to evaluate textual coherence, thematic progression, and pragmatic aspects such as contextual appropriateness and implicit meaning.
Cognitive-linguistic modeling
By implementing principles from cognitive linguistics, the detector analyzes how the text reflects human cognitive processes. This includes evaluating conceptual metaphor usage, frame semantics, and how well the text aligns with principles of natural information processing and memory constraints characteristic of human writing.
Discourse analysis and rhetorical structure
The detector applies discourse analysis theories and Rhetorical Structure Theory (RST) to examine the overall organization of the text. It analyzes cohesion markers, discourse particles, and rhetorical relations between text segments to assess whether the text demonstrates natural structure and discourse coherence.
Frequently Asked Questions
How reliable is your AI detector?
Our AI detector uses four different advanced analysis methods: linguistic analysis with statistical patterns, stylometric profiling, semantic and pragmatic analysis, and cognitive-linguistic modeling. By combining these methods, we can provide a comprehensive assessment, but as with all AI detectors, there is no 100% guarantee - the result should always be seen as support for human judgment.
What techniques does the detector use to analyze the text?
The detector performs an in-depth analysis that includes lexical diversity, syntactic complexity, n-gram frequency analyses, and advanced NLP techniques. It creates a stylistic profile by analyzing sentence length, word choice, and stylistic devices, and evaluates the text's semantic coherence and cognitive processes.
Can the detector identify text from all AI models?
The detector is trained to recognize common patterns from AI-generated text, but different AI models have different 'fingerprints'. Newer and more advanced models can sometimes be harder to detect. We continuously update our analysis methods to keep up with developments.
What do the different probability levels mean?
The result is shown as a percentage where 80-100% means 'Most likely written by an AI', 60-79% means 'Likely written by an AI', 40-59% is 'Uncertain', 20-39% is 'Likely written by a human', and 0-19% is 'Most likely written by a human'. The color coding helps to quickly understand the result.
Can the detector analyze text in languages other than English?
The detector is optimized for English texts as it uses language-specific linguistic models and pattern recognition. For other languages, accuracy may vary because linguistic structures and stylistic characteristics differ between languages.
How should I interpret an 'uncertain' assessment?
When the detector gives an uncertain assessment (40-59%), it means the text contains both human and AI-like characteristics. This may be because the text has been edited by a human after AI generation, or because the author has a very structured writing style. We then recommend further review.
Is my text saved when I use the detector?
All data is processed according to GDPR and stored securely within the EU. The text is analyzed to give you the result and is automatically deleted as soon as the assessment is complete. We never save your text.
Contact Us
Contact us if you have questions, feedback, or simply have a question!