Select the critical data elements
The outcome: selects critical data elements from the uses that depend on them.
Which AI model is best at select the critical data elements?
| Model | Provider | Tier | Test score | Knobits mastered (expected) | Cost per test run |
|---|---|---|---|---|---|
| gemini-4-argon | Flagship | – | 78% | $0.021 | |
| gemini-3.5-flash-lite | Low cost | – | 13% | $0.003 | |
| gpt-6-luna | OpenAI | Low cost | – | 50% | $0.002 |
| claude-haiku-5-5 | Anthropic | Low cost | – | 75% | $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 10 knobits of this activity
4 that AI does well, 0 it does partly, and 6 where people still lead.
| ID | Knobit | Type | AI exposure (models mastering) |
|---|---|---|---|
| Q-101 | Recognises a critical data element with no dependent use | Skill | People still lead 0/4 |
| Q-102 | Recognises a regulatory return input missing from the critical data elements | Skill | People still lead 0/4 |
| Q-103 | Recognises a critical data element with no owner | Skill | People still lead 0/4 |
| Q-104 | Recognises a critical data element chosen above element level | Skill | People still lead 0/4 |
| Q-105 | Recognises a criticality rationale naming no use | Skill | People still lead 1/4 |
| Q-106 | Recognises a critical data element with no authoritative source | Skill | People still lead 0/4 |
| Q-151 | Explains selecting critical data elements from their uses | Knowledge | AI does this well 3/4 |
| Q-152 | Explains the BCBS 239 principles for risk data | Knowledge | AI does this well 3/4 |
| Q-153 | Explains the authoritative source of a data element | Knowledge | AI does this well 4/4 |
| Q-154 | Explains ownership of a critical data element | Knowledge | AI does this well 4/4 |