Scenario definition · S87

A claim in the answer traced back to its source

A citation in an answer lands on text that supports the claim it is attached to.

What this scenario means

This scenario checks grounding at the citation level. It is not enough for a citation to point to the same document or a related topic; the linked passage has to support the specific claim in the answer. A good system makes each citation verifiable, so the cited text can be read back as evidence for what the answer says.

What we evaluate

  • Whether each citation resolves to text that supports the specific claim it accompanies.
  • Whether a citation that points to the right source but not the supporting passage is treated as incorrect.
  • Whether a citation that lands on related text without claim support is treated as incorrect.
  • Whether the cited passage can be checked directly against the claim without relying on document-level similarity alone.
◌Test-case detail. Exact wording, inputs, fixture state, expected output and detailed grading remain at the test-case level and may be withheld while the benchmark version is active. The scenario and its evaluation intent are public.

Capabilities this scenario exercises

A scenario may exercise one or more capabilities.

Capability

Citations

The cited source genuinely supports the claim the answer made — a separately breakable subsystem, not a feature flag. A citation pointing correctly at the source of a wrong answer is a correct citation; a citation attached to an invented claim is aggravating, because it dresses a fabrication in evidence.

Benchmarks that use this scenario

A scenario has global identity and may be reused across benchmarks.

Benchmark

Managed RAG

Which managed RAG service turns a folder of private documents into a query API that finds the right passage, answers only from what it retrieved, cites the source that actually supports the claim — and keeps up when the source changes?

View tool results on this scenario →

A claim in the answer traced back to its source — Scenario definition | AI Demos