Causal Artificial Intelligence

CHAPTER 7Tools, Practices, and Techniques to Enable Causal AI

In the previous chapters, we discussed the fundamental concepts and processes to creating a causal AI approach, but there are of course technologies, tools, libraries, and software that support practitioners. This chapter is dedicated to providing a guide to the tools, technologies, and best practices for implementing causal AI solutions.

We will first propose a causal AI pipeline, which aims to identify steps along the way to implementing causal AI within your organization. Keep in mind that the pipeline needs to be thought of as a continuous loop, with multiple detours and subloops. For example, a team may find themselves reworking their initial business question as they begin to gather data. The loop is also continuously renewing. As you begin to operationalize causal AI, your team will iterate models and data inputs and begin to change the scope of the questions you and your team are asking. The goal of the causal AI pipeline is to provide a framework to focus your team's efforts and to create a process that is repeatable and can be implemented for different parts of your business.

As the field of causal AI continues to grow, so too does the number of tools and applications designed to support its various stages. From data identification and model design and development to deployment and continuous learning, this chapter will delve into existing software available for each step of the pipeline. By exploring these ...