Evaluating AI outputs

Evaluating AI quality uses metrics like BLEU (translation), perplexity (fluency), and human evaluation. No single metric captures everything.

• Multiple metrics evaluate AI output quality.
• Human evaluation remains the gold standard.