Goodhart's Law AI

Definitions

1. The application of Goodhart's Law to AI systems, where optimizing heavily for a measurable proxy can degrade the true objective once the proxy becomes the target. This is a recurring concern in reinforcement learning, ranking, and evaluation design.

   In plain English: When an AI system over-optimizes a metric and stops serving the real goal.

   Example: "Their recommendation system hit a Goodhart's Law AI problem when maximizing click-through rate started surfacing low-quality content that kept users engaged for the wrong reasons."

Related Terms

• Sycophancy AI
• Alignment Tax
• Constitutional AI Method
• Goal Misgeneralization
• Hallucination Rate
• Human Preference
• Mesa-Optimization
• Model Alignment
• Power-Seeking
• Self-Critique
• Specification Gaming
• Uncertainty Quantification
• Validation Metric