S1E10. Can AI Ask Why? The Final Frontier of Causal Reasoning

S1E10. Can AI Ask Why? The Final Frontier of Causal Reasoning

Season 1: The Book of Why As Season 1 comes to a close, we explore the convergence of big data, artificial intelligence, and the age-old question of “why.” While machines have become astonishingly good at pattern recognition, they still struggle with the essence of human understanding: causal reasoning. In this episode, we reflect on how the Causal Revolution challenged traditional statistics and ask whether AI can ever truly emulate human curiosity and imagination. What does it take for machines to not only predict outcomes but explain them? Can we teach AI to distinguish correlation from causation—or even reason about counterfactuals? Join us for a thought-provoking finale as we examine the future of causal thinking in a world increasingly driven by data and algorithms. 🔍 Stay Connected 📧 Email: ⁠⁠⁠⁠amir.rafe@usu.edu⁠⁠⁠⁠ 🌐 Website: ⁠⁠⁠⁠https://pozapas.github.io/⁠⁠⁠⁠ [https://pozapas.github.io/] 🔗 LinkedIn: ⁠⁠⁠⁠https://www.linkedin.com/in/amir-rafe-08770854/⁠⁠⁠⁠ [https://www.linkedin.com/in/amir-rafe-08770854/] 🐦 X: ⁠⁠⁠⁠https://x.com/rafeamir⁠⁠ [https://x.com/rafeamir]

S1E09. Tracing the Invisible: How Causes Travel Through the World

Season 1: The Book of Why What lies between cause and effect? In this episode, we delve into the concept of mediation, the hidden pathways that connect actions to outcomes. From James Lind’s battle with scurvy to groundbreaking diagrams in intelligence research, we explore how scientists uncover the mechanisms that explain how and why effects occur. When we ask, “Does Drug B prevent heart attacks?”, we’re really asking: “Through what chain of events?” Understanding these chains can mean the difference between truth and tragic error. We also reflect on the legacy of Barbara Burks—a pioneering woman in science who challenged the norms of her time by visualizing the complex threads of nature and nurture through path diagrams, decades ahead of her field. Why do some causes work directly while others weave through mediators? And how can this knowledge transform scientific policy, medicine, and artificial intelligence? Join us as we climb deeper into the Ladder of Causation and learn to trace the steps between action and consequence. 🔍 Stay Connected 📧 Email: ⁠⁠⁠amir.rafe@usu.edu⁠⁠⁠ 🌐 Website: ⁠⁠⁠https://pozapas.github.io/⁠⁠⁠ [https://pozapas.github.io/] 🔗 LinkedIn: ⁠⁠⁠https://www.linkedin.com/in/amir-rafe-08770854/⁠⁠⁠ [https://www.linkedin.com/in/amir-rafe-08770854/] 🐦 X: ⁠⁠⁠https://x.com/rafeamir⁠⁠ [https://x.com/rafeamir]

S1E08. Imagining the Impossible: How Counterfactuals Shape Our World

Season 1: The Book of Why What if Cleopatra’s nose had been shorter? What if Joe had taken the aspirin? In this episode, we climb to the top rung of the Ladder of Causation and explore the fascinating world of counterfactuals, alternate realities that help us understand what is and what could have been. We’ll examine how imagining different scenarios is more than philosophical musing, it's central to assigning blame, making predictions, and defining responsibility. From climate change attribution to legal causation, we uncover how counterfactual reasoning powers everything from scientific discovery to everyday decision-making. We also compare two powerful approaches to counterfactuals: the structural causal models that give us a precise computational framework, and the potential outcomes model rooted in statistics. What makes one more effective than the other? And how can machines learn to think in “what ifs” the way humans do? 🔍 Stay Connected 📧 Email: ⁠⁠amir.rafe@usu.edu⁠⁠ 🌐 Website: ⁠⁠https://pozapas.github.io/⁠⁠ [https://pozapas.github.io/] 🔗 LinkedIn: ⁠⁠https://www.linkedin.com/in/amir-rafe-08770854/⁠⁠ [https://www.linkedin.com/in/amir-rafe-08770854/] 🐦 X: ⁠⁠https://x.com/rafeamir⁠⁠ [https://x.com/rafeamir]

S1E07. Climbing the Causal Mountain: How to Predict the Impact of Interventions

Season 1: The Book of Why Understanding what happens when we take action—rather than just observe—is at the heart of causal reasoning. In this episode, we ascend to the second rung of the Ladder of Causation: intervention. How can we predict the effects of a new drug, a change in policy, or even a personal decision if we’ve never observed it before? We explore the tools that make this possible: back-door and front-door adjustments, instrumental variables, and the powerful do-calculus. Why are randomized controlled trials celebrated as the “gold standard”? When can observational data be just as good—or even better? And what happens when we don’t have access to all the confounders? From elegant theory to practical tools, we reveal how modern causal diagrams and equations let us do the unthinkable: predict the consequences of actions not yet taken. 🔍 Stay Connected 📧 Email: ⁠⁠amir.rafe@usu.edu⁠⁠ 🌐 Website: ⁠⁠https://pozapas.github.io/⁠⁠ [https://pozapas.github.io/] 🔗 LinkedIn: ⁠⁠https://www.linkedin.com/in/amir-rafe-08770854/⁠⁠ [https://www.linkedin.com/in/amir-rafe-08770854/] 🐦 X: ⁠⁠https://x.com/rafeamir⁠⁠ [https://x.com/rafeamir]

S1E06. The Paradox Problem: When Data Misleads Our Intuition

Season 1: The Book of Why Sometimes, the numbers lie—or at least, they seem to. In this episode, we dive into some of the most famous paradoxes in statistics and probability, from Simpson’s paradox to the Monty Hall problem. These puzzles reveal the hidden tensions between correlation and causation, challenging our intuition and exposing the limitations of traditional data analysis. Why do paradoxes arise? How can they mislead us in decision-making, science, and AI? And what does it take to resolve them? Join us as we unravel the paradoxes that have baffled researchers for decades and explore how causal reasoning provides the key to seeing through the illusion. 🔍 Stay Connected 📧 Email: ⁠⁠amir.rafe@usu.edu⁠⁠ 🌐 Website: ⁠⁠https://pozapas.github.io/⁠⁠ [https://pozapas.github.io/] 🔗 LinkedIn: ⁠⁠https://www.linkedin.com/in/amir-rafe-08770854/⁠⁠ [https://www.linkedin.com/in/amir-rafe-08770854/] 🐦 X: ⁠⁠https://x.com/rafeamir⁠⁠ [https://x.com/rafeamir]