Analyse the root causes
The outcome: traces data quality failures to the process causes that produce them.
Builds on: Profile and assess data quality.
Which AI model is best at analyse the root causes?
| Model | Provider | Tier | Test score | Knobits mastered (expected) | Cost per test run |
|---|---|---|---|---|---|
| gemini-4-argon | Flagship | – | 88% | $0.021 | |
| gemini-3.5-flash-lite | Low cost | – | 78% | $0.003 | |
| gpt-6-luna | OpenAI | Low cost | – | 88% | $0.002 |
| claude-haiku-5-5 | Anthropic | Low cost | – | 94% | $0.006 |
Test score: mean rubric score on the test tasks for this activity. Knobits mastered: expected share of the activity's knobits the model masters. Cost: mean API cost of one test run at list prices, half price for answers sent as a batch. Method and caveats.
The 9 knobits of this activity
4 that AI does well, 0 it does partly, and 5 where people still lead.
| ID | Knobit | Type | AI exposure (models mastering) |
|---|---|---|---|
| Q-501 | Recognises a data fix applied at the symptom | Skill | People still lead 0/4 |
| Q-502 | Recognises user error given as a root cause | Skill | People still lead 0/4 |
| Q-503 | Recognises one root cause claimed for unrelated failures | Skill | People still lead 0/4 |
| Q-504 | Recognises a root cause the failure distribution contradicts | Skill | AI does this well 3/4 |
| Q-505 | Recognises a root cause fix with no control against recurrence | Skill | People still lead 1/4 |
| Q-506 | Recognises a root cause not looked for elsewhere | Skill | People still lead 0/4 |
| Q-551 | Explains root cause analysis techniques | Knowledge | AI does this well 3/4 |
| Q-552 | Explains corrective and preventive action on a data nonconformity | Knowledge | AI does this well 4/4 |
| Q-553 | Explains testing a root cause against the evidence | Knowledge | AI does this well 3/4 |