(NewsNation) — Anticipatory Policing: Preventing crimes before they occur.
Traditionally confined to the realm of science fiction as depicted in the movie “Minority Report,” the notion of pre-crime intervention is no longer just a futuristic fantasy. Unlike in the film where psychics forewarn authorities of impending murders, contemporary experts assert that the utilization of artificial intelligence can facilitate such predictive capabilities sans the need for clairvoyants.
While the cinematic narrative unfolds in the year 2054, current advancements suggest that pre-crime forecasts are within reach through the application of artificial intelligence algorithms.
In a recent development in 2022, researchers at the University of Chicago harnessed data analytics to devise an algorithm capable of preempting crimes up to a week in advance. By analyzing three years’ worth of data encompassing violent offenses like homicides and assaults, as well as property crimes such as burglaries and car thefts, the algorithm segmented the city into grids approximately a thousand feet apart. Subsequently, it identified hotspots based on the patterns of recent criminal activities.
Upon revisiting the police reports a week later, the researchers discovered that 90% of the city’s homicides, assaults, and thefts transpired within the hotspots identified by the algorithm. This success was replicated when the model was tested in seven other urban areas.
Moreover, beyond mere prediction, the algorithm also aids in identifying individuals with a propensity for involvement in violent crimes, either as perpetrators or victims. By compiling a roster of individuals who frequented the algorithm’s identified hotspots and shared demographic characteristics with past offenders and victims, law enforcement in Chicago leveraged this tool for proactive crime prevention.
Expanding on this concept, other AI-driven crime prediction models delve deeper into assessing the likelihood of recidivism among inmates. By factoring in variables such as prior criminal records, educational background, and employment history, these algorithms gauge an individual’s predisposition to commit future offenses and even speculate on the nature of these crimes.
The envisioned amalgamation of these predictive models aims to not only pinpoint the location of potential crimes but also anticipate the identities of both perpetrators and victims. While the theoretical framework appears promising, widespread implementation faces a significant obstacle—human intervention.
The algorithms, reliant on data derived from human behavior, inadvertently inherit the biases and limitations inherent in human cognition. The identified crime hotspots predominantly centered on black and Latino neighborhoods, underscoring the algorithmic predisposition towards certain demographics. Similarly, the algorithms utilized in parole determinations exhibited a higher tendency to misclassify white inmates as low risk compared to their counterparts of color.
Despite these inherent challenges, proponents of AI-driven crime prevention advocate for remedying biased data through increased data volume. By furnishing AI systems with more comprehensive datasets, there exists a potential for mitigating prejudices and rectifying cognitive fallacies embedded within the algorithms.
However, achieving this level of refinement may necessitate a protracted timeline, possibly spanning decades, with the prospect of realizing comprehensive reform by the projected year 2054.
