59 / 64
The algorithm proved itself by competing against human crime
analysts. In the first phase of the test, human analysts were given
a map of the entire police district every day for 117 days. They
were asked to indicate where crimes were most likely to occur
within a specified 12-hour period.The crime-fighting algorithm
managed to predict the locations of crimes more than twice as
often as the human analysts. In the real-world application of the
study, police officers were dispatched on random days to patrol
areas selected by either the human analysts or the algorithm.
The test was blind, meaning neither the patrol officers nor their
commanders knew whether their orders came from a human
analyst or the computer algorithm. As in the first phase of the
study, the mathematical algorithm resulted in the reduction of
more than twice as many crimes as the human analysts. In ad-
dition to lowering crime, the study suggested that the use of the
predictive mathematical algorithm could save Los Angeles $9
million per year in court, victim, and societal costs.
Researchers at theUniversity of Illinois at Chicago have stretched
the boundary of what a predictive mathematical algorithm can
do by creating one that can interpret what you
intended
to do and
taking its own corrective action. For example, stroke patients of-
ten have to struggle against their own bodies to complete certain
tasks.The algorithmdeveloped inChicago seeks to overcome that
problem by analyzing a person’s actions and determining their
intention.The study hopes that the algorithm can power what it
dubs a “psychic robot,” or a machine that helps complete a task
calculated to be the original owner’s intent. In the case of the
stroke patient, this could translate into a prosthetic that helps
reduce or eliminate shakes or tremors if a patient is intending to
Math and Information Technology
59