1. Definition
Unexamined AI Answers is the specific failure mode where a user accepts and integrates an algorithmic response without interrogating its underlying assumptions, primary evidence, or conceptual framing.
2. Use Case
Activated as a diagnostic warning when a user immediately utilizes a generated fact, code snippet, or argument without attempting to dissect the premise that produced it.
3. Human Role
The learner must notice their passive ingestion, interrupt the seamless copy-paste reflex, and reclaim the forensic role of dissecting how and why the machine arrived at that specific conclusion.
4. AI Role
The system exposes this pattern by occasionally omitting primary sources, providing logically fragile premises, or utilizing an Epistemic Mirror to force the user to identify the unstated assumptions.
5. Friction
The interruption mechanism involves structural delays, demanding that the user explicitly cite an independent source or verbally reconstruct the causal chain before the answer can be formally integrated.
6. Risk
If this pattern continues, the user builds their knowledge base on unquestioned and potentially hallucinated foundations, leaving them vulnerable to severe logical errors when the context shifts.
7. Observable Markers
Recovery is signaled when the user actively challenges the AI’s response (e.g., “What evidence supports this specific framing?”), or manually tests the code/logic before accepting the output.