# Golden Dataset

> A curated set of representative examples and expected outcomes used to evaluate an AI system.

- Category: Evaluation & Safety
- Stage: Established
- Also called: Gold set, Reference dataset
- Canonical: https://www.joelcsaji.com/ai-dictionary/golden-dataset

## Plain English

It is the answer key for the cases you care about most.

## Example

A medical intake assistant is tested against reviewed patient scenarios before each release.

## Why it matters

Without a trusted dataset, teams can confuse demo quality with real improvement.

## Related

- [Evals](https://www.joelcsaji.com/ai-dictionary/evals)
- [Judge Model](https://www.joelcsaji.com/ai-dictionary/judge-model)
- [Trace Grading](https://www.joelcsaji.com/ai-dictionary/trace-grading)

## Sources

- [OpenAI: Trace Grading](https://platform.openai.com/docs/guides/trace-grading)
- [Stanford HAI: 2026 AI Index](https://hai.stanford.edu/ai-index/2026-ai-index-report)