Artificial intelligence (AI) has permeated nearly every facet of modern life, including law enforcement. In the world of policing, AI holds the promise of making investigations more efficient and effective. However, as AI technologies gain traction in the field, concerns about bias and ethical considerations have emerged.
Challenges in facial recognition
One of the most visible applications of AI in policing is facial recognition technology. Police forces have turned to AI algorithms to sift through vast amounts of CCTV and image data, reducing the time and cost required for face identification. However, growing evidence suggests that these systems are far from infallible. AI facial recognition is prone to gender and racial bias, with a pronounced lack of accuracy when it comes to young women of color.
The root of the problem lies in the training data. AI algorithms learn from the faces they are exposed to during their training phase, and if this data predominantly consists of one demographic, it can lead to skewed results. Consequently, individuals from underrepresented groups may be more likely to be misidentified.
Another area where AI has made inroads is predictive policing. Algorithms analyze historical data to predict where and when crimes occur or identify potential offenders. While this approach seems promising, early studies have raised red flags.
Predictive policing relies on historical crime data, often riddled with biases. The result is that AI models may inadvertently label individuals from marginalized communities as disproportionately “dangerous” or “lawless.” For instance, a study in 2016 revealed that Chicago’s “heat map” for anticipated violent crime led to more arrests in low-income or diverse neighborhoods without a corresponding reduction in gun violence. This has prompted EU policymakers to introduce regulations, including a ban on predictive policing software.
Forensic advancements and AI
On a more positive note, AI is proving invaluable in forensics. Complex data, such as DNA analysis and digital evidence, can be overwhelming for human experts. AI can swiftly process this data, making it a powerful tool for investigators.
Recent developments like PACE, an AI image analysis system, can count microscopic particles like pollen or gunshot residue on suspects’ shoes. Counting such particles manually would take months for a human forensics expert, but AI can accomplish it in hours. Additionally, AI-driven analytics streamline the handling of large datasets, such as bank or phone records, aiding investigators in identifying clues and connections rapidly.
While AI’s capabilities are impressive, they fall short in understanding criminal behavior’s emotional and irrational aspects. Many crimes are driven by powerful emotions like anger, hatred, greed, and fear, which AI struggles to comprehend. This underscores the indispensable role of human judgment and intuition in solving complex cases.
Public trust and ethical considerations
The growing integration of AI into policing has raised ethical questions and concerns regarding bias, accountability, and transparency. Policymakers recognize the need for guidelines to ensure AI technologies are used ethically and responsibly. The UK Parliament’s Horizon Scanning Report assesses the impact of AI on policing as “high” and estimates that these changes may occur within the next five years.
In response to the challenges posed by AI in law enforcement, the Home Office has recently established an “AI Covenant” in collaboration with the National Police Chiefs Council (NPCC). This covenant outlines principles to guide the ethical use of AI in policing, aiming to strike a balance between efficiency and fairness.