Recognises binary floating point used for an exact quantity
FLOAT, REAL or DOUBLE for money, areas, shares or anything compared for equality or totalled. Totals do not tie and equality tests fail. A specific form of row 17.
Can AI do this?
AI does this well Most of the tested AI models master this. 10 of 10 models master it.
| Model | Tier | Probability of mastery | Status |
|---|---|---|---|
| gemini-4-argon | Flagship | 99% | Mastered |
| gemini-3.1-pro-preview | Flagship | 99% | Mastered |
| gpt-6-astra | Flagship | 100% | Mastered |
| claude-fable-5-1 | Flagship | 99% | Mastered |
| gemini-3.8-flash | Mid tier | 99% | Mastered |
| gpt-6.1-sol | Mid tier | 100% | Mastered |
| claude-sonnet-5-5 | Mid tier | 100% | Mastered |
| gemini-3.5-flash-lite | Low cost | 99% | Mastered |
| gpt-6-luna | Low cost | 100% | Mastered |
| claude-haiku-5-5 | Low cost | 100% | 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: The column and the exact decimal type it needs
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: payment_amount FLOAT: the sum over payments differs from the statement total by cents. loan_interest_rate REAL: a repayment schedule rebuilt from it does not reproduce the instalments Acceptable: NUMERIC(12,2) for euro amounts, NUMERIC(9,6) for interest rates, scale as the business states it; floating point only for quantities that are approximate by nature (a model's risk score)
Also known as: Approximate number for an exact quantity (physical model); Money in binary floating point (generated DDL)
Related knobits
Builds on
All 40 knobits of Design the physical data model.
Source: Review checklist 06-physical, row 18; Karwin, Rounding Errors; S&W 11.4