Recognises a root cause the failure distribution contradicts
A cause that predicts where the failure should occur, contradicted by where it does.
Can AI do this?
AI does this well Most of the tested AI models master this. 3 of 4 models in the current benchmark master it.
| Model | Tier | Probability of mastery | Status |
|---|---|---|---|
| gemini-4-argon | Flagship | 99% | Mastered |
| gemini-3.5-flash-lite | Low cost | 46% | Not shown |
| gpt-6-luna | Low cost | 99% | Mastered |
| claude-haiku-5-5 | Low cost | 99% | Mastered |
Mastered means a probability of at least 90% that the model has this knobit, estimated from its graded answers. “Not shown” means the answers do not settle it either way. How the models are tested.
What it looks like in practice
Expected result: A cause consistent with where the failure occurs and where it does not
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: CA-004 blames app onboarding for employer "-"; branch rate 6.7% vs app 6.9% Acceptable: Both channels share the onboarding service, which accepts "-"
Also known as: Cause the evidence does not support
Related knobits
Builds on
All 9 knobits of Analyse the root causes.
Source: Review checklist 05-root-cause-analysis, row 4 (9001 10.2.1(b); DMBOK)