AI Metric

Definitions

1. A measurement used to assess some aspect of AI system behavior, such as accuracy, latency, cost, hallucination rate, acceptance rate, or user satisfaction. AI metrics help teams track progress, but no single metric usually captures overall usefulness.

   In plain English: A measurement used to judge how an AI system is performing.

   Example: "They added a new AI metric for citation correctness because raw thumbs-up rates were hiding factual failures."

Related Terms

• Perplexity Score
• AI Performance
• AI Quality
• AI Testing
• Hallucination Rate
• Human Baseline
• LLM Benchmark
• Model Eval
• Prompt Evaluation
• Safety Evaluation
• Token Count
• Truthfulness
• Uncertainty Quantification
• Validation Metric