Causal AI: A Theoretical Introduction

In this course, you'll learn the foundational components of Causal Artificial Intelligence (Causal AI) / Causal Inference.

More and more people are starting to realise that correlation-focused models are not enough to answer our most important business questions. Business decision-making is all about understanding the effect different decisions have on outcomes, and choosing the best option. We can't understand the effect decisions have on outcomes with just correlations; we must understand cause and effect.

Unfortunately, there is a huge gap of knowledge in causal techniques among people working in the data & statistics industry. This means that causal problems are often approached with correlation-focused models, which results in sub-optimal or even poor solutions.

In recent years, the field of Causality has evolved significantly, particularly due to the work of Judea Pearl. Judea Pearl has created a framework that provides clear and general methods we can use to understand causality and estimate causal effects using observational data. Combining his work with advances in AI has given rise to the field of Causal Artificial Intelligence.

Causal AI is all about estimating causal effects (using observational data). Generally, businesses rely only on experimentation methods like Randomized Controlled Trials (RCTs) and A/B tests to determine causal effects. Causal AI now adds to this by offering tools to estimate causal effects using observational data, which is more commonly available in business settings. This is particularly valuable when experimentation is not feasible or practical, making it a powerful tool for businesses looking to use their existing data for decision-making.

This course is designed to bridge the knowledge gap in causal techniques for individuals interested in data and statistics. You will learn the foundational components of Causal AI, with a specific focus on the Pearlian Framework. Key concepts covered include The Ladder of Causation, Causal Graphs, Do-calculus, and Structural Causal Models. Additionally, the course will go over various estimation techniques, incorporating both machine learning and propensity score-based estimators. Last, you'll learn about methods we can use to obtain Causal Graphs, a process called Causal Discovery. The course has a theoretical focus and provides the foundation needed to get started in Causal Inference.

By the end of this course, you'll be fully equipped with knowledge on what it takes to estimate average causal effects using observational data. 

We believe that everyone working in the data and statistics field should understand causality and be equipped with causal techniques. By educating yourself early in this area, you will set yourself apart from others in the field. If you have a basic understanding of probability and statistics and are interested in learning about Causal AI, this course is perfect for you!