AI · Flashcard
Asked whether it is sure, Claude changes its answer. What has that change actually demonstrated?
- AThat the model reacted to pressure in the prompt, which is no evidence that either answer is right
- BThat the first answer was wrong, since a correct answer would have survived the challenge intact
- CThat the second answer is verified, since re-checking triggers a stricter internal review pass
- DThat confidence was low the first time, which the model reveals by revising when it is questioned
Why this is the answer
A challenge in the prompt is itself an input, and the model can revise a correct answer or defend a wrong one — which is why grounding the answer in a source, rather than re-interrogating the model, is the recommended way to reduce fabrication. Revision does not identify which answer was wrong, no stricter review pass is triggered by the question, and there is no confidence level being disclosed.
Official docsMore Claude Certified Associate — Foundations flashcards
- A researcher asks Claude for sources and gets three URLs, two of which lead to 404 pages. What explains the dead links?
- A manager worries that connecting a document store to Claude will expose files their team cannot normally open. What actually governs access?
- A firm's risk officer asks whether work done on a commercial Claude plan is used to train Anthropic's models. What is the documented default?