This repository compiles high-profile real-world examples of failed machine learning projects, serving as a learning resource for engineers, researchers, and practitioners. It highlights common pitfalls and biases across various ML domains, offering valuable insights to prevent similar failures in future applications.

How It Works

The project functions as a curated collection of case studies, categorizing failures by ML domain: Classic ML, Computer Vision, Forecasting, Image Generation, Natural Language Processing, and Recommendation Systems. Each entry details a specific project, its description, and the nature of its failure, providing concrete examples of bias, inaccuracy, or unintended consequences.

Quick Start & Requirements

No installation or execution is required. This is a static collection of information.

Highlighted Details

• Features over 30 documented case studies of ML project failures.
• Covers a wide spectrum of ML applications, from recruitment systems and facial recognition to chatbots and financial trading.
• Includes examples of bias (racial, gender), factual inaccuracies, data leakage, and unexpected emergent behaviors.
• Provides context on the impact of these failures, ranging from reputational damage to regulatory investigations.

Maintenance & Community

This is a static, curated list. There is no active development or community interaction indicated.

Licensing & Compatibility

The repository content is not explicitly licensed.

Limitations & Caveats

The repository is a passive collection of information and does not offer any tools, code, or active analysis. The scope is limited to publicly documented failures.