Self-driving car systems trained primarily on data from sunny California highways have shown reduced reliability when deployed in snow, fog, or unfamiliar road layouts. Researchers have addressed this in part by deliberately incorporating diverse weather and geographic conditions into training data. Even so, edge cases — such as pedestrians in costumes or unusual construction zones — continue to pose challenges that the systems sometimes mishandle.
Based on the passage, which inference is most strongly supported?
- Acheck_circle
The performance of trained systems can depend on how well their training inputs cover the situations they encounter
- B
Diverse training data eliminates all performance limitations
- C
Pedestrians in costumes are the primary safety concern in autonomous driving
- D
Self-driving systems perform identically in all conditions regardless of training
Explanation
Performance varying with training-data coverage and ongoing edge-case difficulty supports A. B, C, D contradict the passage.