Predicting outcomes has always been a big business and valuable for a variety of reasons. With new technology and sophisticated algorithms the prediction business is thriving. Large retailers can determine with reasonable accuracy what consumers will want, how much of it they'll want and what they'll pay to get it. Algorithms can predict many of the aftershocks after an earthquake. Trading algorithms can give investment companies a head start on where major stock markets are heading.
Using similar technology to predict where and when a crime will occur, however, seems to be in the realm of science fiction.
Yet several police departments around the country are buying into the idea. Jeff Brantingham, a UCLA anthropologist, along with other social scientists and mathematicians, recently created "PredPol," a computer software program that claims to predict where and when crime is likely to occur.
On its first day of implementation in the Norcross Police Department, for example, officers used PredPol to apprehend a fugitive and two burglars. The department credits PredPol as helpful in patrolling the areas where the suspects were arrested. The Atlanta Police Department has already initiated a pilot program of PredPol this summer and may implement a similar program to the one used in Norcross sometime this fall.
The software does not claim to predict who will commit a crime. Rather it gives officers "predictive boxes" of times and places where the algorithm indicates a crime is likely to occur. The area may be as small as a 500-by-500-foot square box. In Norcross, police use it to track burglaries and thefts. However, other cities have used it more comprehensively. Seattle was the first city to use the program in an attempt to predict gun violence.
Legal scholars and criminal defense attorneys have already raised questions and concerns regarding the use of this technology. Police departments have so far told officers not to use PredPol as a basis for stops or searches. It is highly questionable whether an algorithm amounts to a "reasonable or articulable suspicion" that someone is behaving criminally.
An officer needs an articulable suspicion to pull over or stop and search someone, which is a grey area of the law criminal courts constantly scrutinize. At this point, it is unclear whether an officer will testify to the use of a computer model in court when articulating his or her reasonable suspicion. It is also unclear whether the use of PredPol in apprehending a suspect may be omitted in an officer's testimony.
Courts have not yet weighed in on whether an officer's conducting a search only using a computer model is Constitutional. It is doubtful this would be deemed to have sufficient particularization.to be upheld. The issue may soon arise in courts across the country. Some argue that using an algorithm to patrol an area contains less biased decision-making than other methods, since a computer cannot racially profile. However, what data goes into the algorithm may itself be biased and lead to even greater profiling, an issue that has plagued Georgia and other states for decades.
People in Georgia accused of criminal activity should contact an experienced criminal defense attorney to ensure their constitutional rights are upheld in court.