AI & Climate Quiz

Answers and explanations appear as you take the quiz.

1. What information should accompany a climate-related prediction?
   • Location, time horizon, uncertainty, and intended decision
   • Only the model name
   • A single number without context
   • The number of training epochs
2. Why should climate models be tested on unusual events as well as average conditions?
   • Average performance guarantees extreme-event performance
   • Unusual events should always be removed
   • Rare events may matter most for planning even if average error is low
   • Rare events have no effect on decisions
3. What can measurement changes do to an apparent environmental trend?
   • Guarantee the trend is real
   • Create an apparent change unrelated to the climate outcome
   • Remove missing observations
   • Make geographic coverage equal
4. How should model performance transfer to a different climate or sensor setup be treated?
   • Assume it works if the model is large
   • Use the same score without checking
   • Reuse the original aggregate score as proof of local performance
   • It needs evaluation because transfer is not automatic
5. What should a team measure when considering AI’s climate impact?
   • Compute use and downstream operational effects under stated assumptions
   • Only the number of parameters
   • Only the model’s predicted score
   • The number of data points
6. What does a model’s compute estimate establish about a full lifecycle impact?
   • It equals every downstream effect
   • It proves emissions fell
   • It is one component, not automatically a complete lifecycle assessment
   • It establishes a universal carbon saving
7. Why check geographic coverage and measurement quality before using a climate forecast in a community?
   • Coverage does not affect climate applications
   • The available observations and evaluation may not represent that location or its conditions
   • It guarantees every location has sensors
   • It removes the need for uncertainty
8. Who should retain authority over environmental or emergency actions?
   • The model automatically
   • The sensor vendor
   • The training dataset
   • The accountable decision-maker using evidence and review