Enterprise Artificial Intelligence Transformation

Chapter 8The AI Lifecycle

Our success at Amazon is a function of how many experiments we do per year, per month, per week, per day …

Jeff Bezos, CEO of Amazon

This chapter provides a high-level AI model creation workflow: a description of the steps needed to execute an AI project at scale, from finding use cases to getting the model deployed and used in production. Some of the steps are iterative and may overlap, but in general, they take place in the order described. Understanding the difference between a variety of machine learning solutions can be challenging for those who are not AI experts, and there may be many algorithms and a variety of ways that can be used to solve business needs. This does not mean you have to become an expert in every modeling nuance to manage an AI team or project successfully. However, it is essential to understand not just what the steps in creating each model are but why they are necessary, allowing team managers to distinguish between what is working and what is not. Additionally, understanding this lifecycle makes clear why the AI platform features described in the next chapter are necessary.

Figure 8.1 shows the high-level end-to-end process for modeling. The first step in the process is to identify and define use cases. For each use case, the team decides what specific questions they want answered and determines how answering these questions benefits the business. It is crucial to resolve decisions early on about how to frame a use case ...