Examining a Pipeline Design for an AI System

When your software system is small, such as a small game that you built for your use, it's easy enough to remember all of the pieces mentally and to have a good understanding of how everything works and how it fits together. For complicated systems, such as HealthCare.gov, which we examined previously, it is very normal for no single person to have a complete understanding of how the whole system works. Instead, a few people or a few teams will understand and be responsible for specific components of a system. For example, the responsibility of a global system might be split by region, with one team responsible for America, one for Europe, and so on. Or, responsibility might be split by function, with one team for machine learning, one for data collection, and others for other functions besides.

Let's imagine a hypothetical system that uses information from news articles to inform the end-user of financial trading decisions...