A government agency is operationalizing an AI system to optimize urban traffic flow that changes...

PMI’s CPMAI-aligned guidance emphasizes that AI initiatives must be managed as continuous lifecycles and that teams must address the gap between models and real-world implementation, including challenges such as changing conditions that can drive performance degradation (e.g., drift). In a traffic optimization use case where conditions change unexpectedly, the governing need is not merely to have more data, but to ensure the AI solution is operating on current, synchronized inputs across multiple data sources (cameras, sensors, historical patterns) so that recommendations reflect the present state of the system. Real-time synchronization directly supports this by aligning timestamps, ensuring consistent ingestion across feeds, and enabling timely analysis for decision-making when traffic conditions shift quickly. This approach best matches the operational objective of optimizing a dynamic environment because it reduces latency and inconsistency between streams, which otherwise can lead to outdated or conflicting interpretations. While data augmentation (B) can help model robustness, and dimensionality reduction (D) can manage complexity, neither guarantees that the operational system is using the most current multi-source view. Therefore, real-time data synchronization is the most effective technique for the stated goal.