Causal Artificial Intelligence [Free Textbook] (causalai-book.net)

Elias Bareinboim has released a comprehensive draft of his free textbook, *Causal Artificial Intelligence: A Roadmap for Building Causally Intelligent Systems*, aimed at researchers and advanced students in AI, machine learning, and statistics. The book unifies fundamental concepts from probability theory, causal inference, and decision-making under uncertainty to address key challenges in AI—such as robustness, generalization, fairness, and explainability—through a causal lens. It offers a rigorous yet accessible treatment of Structural Causal Models, Pearl’s Causal Hierarchy, counterfactual reasoning, and causal decision-making, supplemented by an extensive lecture series and modular course tracks. This textbook is significant for the AI/ML community because it provides both the theoretical foundations and practical algorithms necessary to embed causal reasoning into modern AI systems. By moving beyond correlational models to emphasize cause-effect relationships, it equips practitioners to build more reliable and interpretable models that can safely generalize across environments, make decisions under uncertainty, and address ethical concerns like fairness. Key technical topics covered include algorithmic identification of causal effects, counterfactual calculus, causal reinforcement learning, domain adaptation, causal generative modeling, and advanced topics like parametric identification and causal estimation. Bareinboim’s work represents a pivotal resource for advancing causal AI as a discipline. Its clear structure supports diverse educational pathways, from foundational introductions to cutting-edge research areas. As causality gains traction in the AI community for tackling complex inference and robustness problems, this textbook offers a unified, rigorous roadmap to develop truly causally intelligent systems.