🕸️

Judea Pearl

Original nameJudea Pearl

American Computer Scientist, Founder of Bayesian Networks and Causal Inference Theory

Those Who Conceived That "a Machine Could Think" · Pioneers

Bayesian NetworksCausal Inference TheoryThe Book of Why

Who they are

Judea Pearl (1936– ) is an Israeli-born computer scientist long working in the US who in 1988 published Probabilistic Reasoning in Intelligent Systems, systematically proposing "Bayesian networks," a framework representing the probabilistic dependencies among variables with a graph structure, providing a rigorous mathematical tool for AI systems to reason in an uncertain real world, thereafter widely applied in medical diagnosis, fault troubleshooting, natural language processing, and many fields. Pearl thereafter shifted his focus to "causal inference," proposing the "do-calculus," a mathematical tool for distinguishing "correlation" from "causation," and popularizing this thought in his 2018 book The Book of Why: The New Science of Cause and Effect. He repeatedly stresses that the current mainstream AI systems represented by deep learning are in essence still at the level of "curve fitting," able only to recognize correlation and unable to understand causation, which he holds a key gap remaining on the road to true general artificial intelligence. Pearl won the Turing Award in 2011.

Primary sourcesPearl, Probabilistic Reasoning in Intelligent Systems (1988)Pearl & Mackenzie, The Book of Why (2018)

Key stories

"The Rooster Crows, the Sun Rises" — Correlation Is Not Causation

Pearl often uses a plain example to show the difference of correlation and causation: the rooster crowing is highly correlated with the sun rising, but the rooster crowing is plainly not the cause of the sun rising. He holds that deep-learning systems relying only on vast data to find statistical correlation can in essence hardly escape such traps, and only by introducing an explicit causal structure can a machine truly understand counterfactual reasoning such as "what would happen if I did A," his core reason for keeping a cautious attitude toward the purely data-driven deep-learning path.

Relationships

Mentor

📐Newell and Simon

Ally

📊Vladimir Vapnik　The two, from the two different paths of causal inference and statistical learning theory, shaped the theoretical basis of the AI field’s understanding of "how to learn reliably from data"

Echoes today

Below are how modern works borrow or reinterpret this name or story — not the original material. The two differ, so keep them apart.

The public science of "correlation is not causation"The popular metaphor of "the rooster crows, the sun rises" that Pearl proposed thereafter became one of the most-cited examples for the public to understand the common trap of the "correlation fallacy" in statistics and data science.

Appears in

The Perceptron, the Winter, and the Revival of ConnectionismStatistical Learning Theory and Bayesian Networks

← Back to Figures