Judea Pearl receives the 2011 ACM Turing Award

Background

Much of what an intelligent system needs to know about the world is uncertain. Systems that treat knowledge as strictly true or false run into trouble as soon as evidence is missing or noisy, yet handling uncertainty with probability brings its own difficulty: the number of combinations grows explosively with the number of variables, so a naive model becomes impossible to specify, store or compute with. Judea Pearl's work addressed both sides of that problem, and the 2011 A.M. Turing Award recognised it ^[1].

The 2011 award

ACM lists Judea Pearl as the recipient of the 2011 A.M. Turing Award ^[1]. The citation reads: "For fundamental contributions to artificial intelligence through the development of a calculus for probabilistic and causal reasoning." ^[1] The wording points to two distinct achievements: a calculus for reasoning under uncertainty, and a calculus for reasoning about cause and effect.

Bayesian networks

Pearl invented Bayesian networks, a mathematical formalism for defining complex probability models, together with the principal algorithms for reasoning with them ^[1]. A Bayesian network describes a set of variables as nodes of a graph, with arrows recording which variables directly influence which; each node then needs only a small table of probabilities conditional on its parents, rather than one enormous table covering every combination of values. Because the graph encodes the structure of the problem, the model stays compact and inference can be carried out efficiently, and it can also be updated when new evidence arrives.

His 1988 book "Probabilistic Reasoning in Intelligent Systems" sparked a revolution in the field of artificial intelligence ^[1]. The book made the case that uncertainty should be handled with probability rather than with ad hoc certainty factors and hand-written rules, and it gave researchers a working toolkit of representations and algorithms.

From probability to causation

Probability alone describes how often things happen together, not what happens when someone intervenes. His 2000 book "Causality" presented do-calculus, a complete mathematical framework for building causal models and analyzing causal relationships from data ^[1]. The framework introduces an explicit notation for intervention — asking what the probability of an outcome would be if a variable were deliberately set to a given value — and supplies rules for manipulating such expressions. Those rules let a researcher decide whether a causal effect of interest can be computed from the data actually available, and if so how, rather than assuming that a statistical association is a causal claim.

At UCLA, and the prize money

He was Professor of Computer Science at UCLA and founder and director of its Cognitive Systems Laboratory ^[1]. Beyond the research itself, Pearl announced he would donate a major portion of the Turing Prize money to support the projects of the Daniel Pearl Foundation ^[1].

Why it matters

The award citation names what the two strands share: both are calculi, formal systems in which questions about evidence and about causes can be posed precisely and answered by rule. Bayesian networks gave artificial intelligence a practical way to represent and reason with uncertain knowledge, and the do-calculus gave it a language for distinguishing what is merely observed from what is caused. Together they moved the field from systems that manipulated symbols with fixed certainty toward systems that quantify uncertainty and, beyond that, reason about the effects of actions.