PERSON
Pedro Domingos
The computer scientist who believes the dozens of competing methods in machine learning are five tribes of one field—and who has spent his career arguing, with equal force, that a universal learner is findable and that today’s machines, however impressive, are not it.
Pedro Domingos is the unifier of machine learning—a Lisbon-born computer scientist who organized the entire field around the suspicion that its apparently competing methods are five disguises of one underlying idea. His 2015 book
The Master Algorithm introduced a wide public to the
field’s five tribes—symbolists, connectionists, evolutionaries, Bayesians, and analogizers—and proposed that a single universal learning algorithm, from which all knowledge could be derived from data, awaits discovery. What distinguished Domingos from the public faces of AI is that he held two positions simultaneously that rarely coexist: the most ambitious vision of what machines could eventually learn, and the most deflationary assessment of what today’s systems actually are. His most quoted line—that the danger is not that computers will become too smart but that they are too stupid and have already taken over the world—is a compressed thesis about where the real risks of
automation lie. His technical contributions, from Markov logic