R0054/2026-03-31/C007 — Assessment¶
BLUF¶
The claim is exactly correct. The prompt defines a researcher profile with three sections (biases, COI, blind spots), instructs the AI to use it for transparent calibration at Step 1, and instructs verification against it during self-audit at Step 9. The concept is grounded in established systematic review methodology (Cochrane, NAS) and extends it to AI-researcher interaction.
Probability¶
Rating: Almost certain(ly) / Nearly certain (95-99%)
Confidence in assessment: High
Confidence rationale: Direct verification against primary source. All three elements of the claim (profile content, calibration, audit verification) are explicitly documented.
Reasoning Chain¶
-
FACT: The prompt's Researcher Profile section defines a template with three categories: Declared Biases, Conflicts of Interest, and Acknowledged Blind Spots. [SRC01-E01, High reliability, High relevance]
-
FACT: Step 1 (Receive and Clarify) includes a "Researcher profile check" that instructs the AI to review the profile against the specific input, stop if relevant biases/COI/blind spots are found, and transparently confront the researcher before proceeding. [SRC01-E01, High reliability, High relevance]
-
FACT: Step 9 (Self-Audit) includes instructions to "check the researcher profile" to determine if declared biases or COI influenced questions, searches, or interpretation. [SRC01-E01, High reliability, High relevance]
-
FACT: COI declaration is standard practice in systematic reviews, as documented in the Cochrane Handbook. [SRC02-E01, High reliability, Medium relevance]
-
REPORTED: AI can be used for COI detection and management in research contexts. [SRC03-E01, High reliability, Medium relevance]
Evidence Base Summary¶
| Source | Description | Reliability | Relevance | Key Finding |
|---|---|---|---|---|
| SRC01 | prompt-snapshot.md | High | High | Profile with three sections, used at Steps 1 and 9 |
| SRC02 | Cochrane Handbook | High | Medium | COI declaration is established research methodology |
| SRC03 | PMC COI article | High | Medium | AI-mediated COI management validated |
Collection Synthesis¶
| Dimension | Assessment |
|---|---|
| Evidence quality | Robust — primary source confirmation + established methodology validation |
| Source agreement | High |
| Source independence | High — primary source, Cochrane, and PMC are independent |
| Outliers | None |
Gaps¶
| Missing Evidence | Impact on Assessment |
|---|---|
| No evaluation of whether the researcher profile actually changes AI behavior | Would provide functional proof, not just structural verification |
Researcher Bias Check¶
Declared biases: Researcher is prompt author (COI).
Influence assessment: Minimal for structural verification. The claim is about what the prompt says, not about whether it works effectively.
Cross-References¶
| Entity | ID | File |
|---|---|---|
| Hypotheses | H1, H2, H3 | hypotheses/ |
| Sources | SRC01, SRC02, SRC03 | sources/ |
| ACH Matrix | — | ach-matrix.md |
| Self-Audit | — | self-audit.md |