引導式測驗 · 中等 等級
AI 災難建模測驗
了解災難模型組件以及人工智慧如何支援危害、暴露和損害估計,同時考慮不確定性和治理。
問題 1 的 8
Which components commonly contribute to catastrophe loss modeling?
預覽每個問題8 問題
當您參加測驗時,就會出現答案和解釋。
- Which components commonly contribute to catastrophe loss modeling?
- Only property values
- Hazard and property values, without damage behavior or contract terms
- Claims history and policy limits, without event locations or asset exposure
- An image-classification score without hazard, exposure, or vulnerability data
- What does an AI damage estimate represent?
- A guaranteed claim payment
- An uncertain model output that needs validation
- A precise forecast of the next disaster
- A replacement for policy terms
- Why can claims data bias catastrophe models?
- Claims always cover all losses
- Claims are generated before events
- Bias is removed by adding more model layers
- Claims may be incomplete or reflect which properties were insured and reported
- How can exposure data affect modeled losses?
- Stale or missing asset details can change the estimated affected portfolio
- Exposure matters only for premium pricing, not loss estimation
- Only the model architecture changes which assets are exposed
- A deductible can substitute for knowing what properties lie in the hazard area
- Why test model sensitivity to hazard and vulnerability assumptions?
- All assumptions are known exactly
- Sensitivity analysis proves an event will occur
- Different assumptions can materially change loss estimates
- The test replaces data governance
- How should a catastrophe loss model output be interpreted?
- The model's accuracy on each building image
- The exact date and location of the next disaster
- A building-level guarantee of the claim payment
- A determination that a particular property is safe from future events
- Why may historical data be insufficient for future catastrophe conditions?
- Historical data are always forward-looking
- The model ignores hazard intensity
- Claims labels never change
- Climate, construction and exposure can change over time
- What should accompany an AI-supported insurance decision?
- Validation, documented assumptions and review of customer impact
- Only a high model score
- A claim that the model is objective
- No explanation of data sources