Câu đố về AI trong Hậu cần

Câu trả lời và giải thích xuất hiện khi bạn làm bài kiểm tra.

1. A routing model minimizes distance but misses promised delivery windows. What is the issue?
   • The shortest route always meets every commitment
   • Traffic is irrelevant once distance is minimized
   • The model has no input data
   • The objective omitted an important service constraint
2. Which information can change an estimated arrival time beyond distance?
   • Traffic, weather, and changing route conditions
   • Only the customer’s name
   • Số tham số của mô hình
   • The truck’s color
3. Why test demand forecasts on later periods and unusual conditions?
   • Later data are always easier
   • Spikes and operational changes can reveal failures hidden by historical averages
   • A forecast should only fit its training period
   • Unusual conditions do not affect logistics
4. A route plan has a high efficiency score but violates driver-hour or capacity limits. What should happen?
   • Accept the route because the score is high
   • Allow the model to decide which law to ignore
   • Treat those limits as hard constraints that the objective cannot trade away
   • Update the prediction after delivery
5. What should an ETA communicate when conditions remain uncertain?
   • A guaranteed arrival time
   • Only the shortest possible travel time
   • An estimate that can be updated, not an unsupported promise
   • The route’s optimization score
6. Why verify package identity and destination before changing shipment records?
   • Verification makes routing slower without benefit
   • A mismatch can send an item to the wrong place or associate data with another shipment
   • The model already knows the recipient’s identity
   • Package IDs never contain errors
7. What is a safe fallback when a model or network is unavailable?
   • A manual dispatch path and safe retry behavior
   • Repeat the same command indefinitely
   • Invent an ETA
   • Delete the shipment record
8. Which result best measures whether a logistics optimization helped?
   • Only predicted distance
   • Number of algorithm iterations
   • Model confidence alone
   • Service level, lateness, fuel, and workload compared with a baseline