AI GLOSSARY · SAFETY & TRUST
Evaluation (evals)
Systematic tests that measure how well a model or prompt performs on defined tasks.
Example
Scoring 50 sample answers against a rubric before going live.
Learn it by doing
Reading a definition is a start; using the idea makes it stick.
Related terms
SAFETY & TRUSTHallucinationWhen a model states something false or invented as if it were true.SAFETY & TRUSTGroundingTying a model’s answer to provided sources so claims can be traced and checked.SAFETY & TRUSTBiasSystematic unfairness in outputs, often reflecting gaps or stereotypes in training data.SAFETY & TRUSTGuardrailsRules and checks around a model that block unsafe inputs or outputs.SAFETY & TRUSTHuman in the loopKeeping a person responsible for reviewing or approving AI outputs before they take effect.SAFETY & TRUSTPrompt injectionAn attack where text inside content tricks a model into ignoring its instructions.