Generative AI Leader Practice Question: A financial services company uses a generative AI…

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Confidence indicators

  Why it's wrong here

  Confidence indicators expose the model's certainty score for a prediction, not the provenance of its statements. Source traceability requires retrieved passages to be cited in the output, as in retrieval-augmented generation with attribution. Confidence scores suit risk triage, where uncertain answers are escalated for human review.

• ✗

  Model Cards

  Why it's wrong here

  Model Cards document a model's intended use, training data and evaluation metrics, but they describe the model itself, not individual outputs. Tracing a specific piece of advice to its source needs per-response citation of retrieved documents. Model Cards would be correct for governance documentation accompanying a released model.

• ✗

  Chain-of-thought reasoning

  Why it's wrong here

  Chain-of-thought reasoning exposes intermediate steps in the model's logic, but those steps are generated text, not verified source references. Tracing outputs to specific sources requires retrieval with citation of the underlying documents. Chain-of-thought suits debugging reasoning quality, not provenance auditing.

• ✓

  Grounding (citing sources)

  Why this is correct

  Grounding retrieves authoritative documents and constrains the model to generate responses tied to those retrieved passages, attaching citations to each claim. This satisfies the traceability requirement by linking every output back to a verifiable source.

Go deeper

Related to this question