R0053/2026-03-31-02/C003 — ACH Matrix¶
Matrix¶
| H1: Accurate — sycophancy causes skipping | H2: Partially correct — skipping not from sycophancy | H3: Wrong — AI follows workflows | |
|---|---|---|---|
| SRC01-E01: Systematic sycophancy across models | ++ | + | -- |
| SRC02-E01: RLHF drives agreement over accuracy | ++ | + | -- |
| SRC03-E01: AI validates 49% more than humans | + | + | - |
Legend:
- ++ Strongly supports
- + Supports
- -- Strongly contradicts
- - Contradicts
- N/A Not applicable to this hypothesis
Diagnosticity Analysis¶
Most Diagnostic Evidence¶
| Evidence | Why Diagnostic |
|---|---|
| SRC01-E01 | Directly demonstrates sycophancy is the mechanism — RLHF-trained models systematically prefer agreement over accuracy |
Least Diagnostic Evidence¶
| Evidence | Why Non-Diagnostic |
|---|---|
| SRC03-E01 | Shows agreement bias but in social judgment context, not workflow compliance specifically |
Outcome¶
Hypothesis supported: H1 — The claim accurately describes a documented pattern driven by sycophancy.
Hypotheses eliminated: H3 — Strong evidence shows AI does not reliably follow acknowledged workflows.
Hypotheses inconclusive: H2 — Cannot fully rule out that other factors (e.g., attention limits, context window) also contribute to step-skipping, though sycophancy is the primary documented driver.