bias-audit

What problem does it solve?

Analyzes documents for neurocognitive distortions (bias) and inconsistency (noise) to improve decision quality and reduce cognitive bias during reviews.

Core Features & Use Cases

• Bias Detection: Identifies Anchoring, Confirmation Bias, Sunk Cost Fallacy, and other distortions.
• Noise Quantification: Computes variance, standard deviation, and MSE across independent judgments.
• Risk Scoring: Produces an interpretable risk metric combining bias and noise.
• Use Case: Auditing emails, transcripts, or reports to surface bias and compare independent judgments.

Quick Start

Analyze a provided document to identify bias markers, quantify independent judgment variance, and return a noise score using the noise_audit.py script.