Calvin Lab Auditorium
Science has struggled for decades to make sense of causation, even though the goals of science are usually to understand causes and effects. We want to know whether smoking causes cancer, or whether exercise will prevent heart disease. Yet the American Medical Association even forbids researchers to use the language of cause and effect unless they have conducted a "randomized clinical trial" – which is not always ethical or possible.
Likewise, machine learning has a blind spot where causation is concerned. A machine cannot tell whether the rooster's crow causes the sun to rise, or vice versa. It cannot even make sense of the question. Before we can make machines or bots or AIs that are as smart as humans, we need to make them as smart as three-year-olds, and we haven't done that yet! Human three-year-olds have a very sophisticated understanding of causation.
In this public lecture, Dana Mackenzie (currently journalist in residence at the Simons Institute) will read from his new book, The Book of Why, which he wrote in collaboration with computer scientist and Turing Award winner Judea Pearl. Mackenzie will explain how Pearl has restored the lost language of causation to science, and placed the study of cause and effect on a firm mathematical foundation. This "causal revolution" has already started to affect medical research, and is poised to transform machine learning as well.
Copies of The Book of Why will be available for purchase at the event.