Lab 1 — Prompt and Verification

Fact-check status

Reviewed claim by claim on 2026-09-02 against the primary or official sources identified inline as S1 and S2. Recommendations are design guidance unless a cited standard makes them normative.

Key jargon

Term	Plain-language meaning
Hypothesis	A testable expectation about how a prompt change will affect results. S1
Test case	One defined input with expected properties or outcome. S1
Control prompt	The unchanged baseline used for comparison. S1
Failure taxonomy	Named categories used to classify observed errors. S1

Key concepts

Change one prompt element at a time and compare against a stable baseline. S1
Record inputs, model/version, settings, output, evidence, and failure labels so results can be reproduced. S1

Concept map

flowchart LR
    A["Define hypothesis and cases"] --> B["Run baseline and variant"]
    B --> C["Verify claims and classify failures"]
    C --> D["Write evidence-backed conclusion"]

Diagram semantics verified against S1 and S2 on 2026-09-02.

Goal

Produce a one-page comparison using three supplied authoritative sources and demonstrate which source supports each material claim.

Steps

Define audience, decision, scope, date, and output contract. S1
Supply three short documents plus source metadata. S1
Require claim-level citations and explicit unknowns. S1
Run three times; record model/version/settings. S1
Open every citation and label support as full, partial, or absent. S1
Revise the prompt once and compare failure classes. S1

Pass criteria

All material factual claims have direct support.
Conflicting evidence remains visible.
No invented source or unavailable fact.
Output format is stable enough for the user.

Artifacts

Task contract, source manifest, prompt versions, outputs, claim audit, decision, and lessons learned.