With the growing recognition of the opportunities for artificial intelligence (AI) to significantly reduce the time and costs associated with some policing activities, the impetus for wider use of this technology is notable.

In the latest in a series of articles on AI in policing, Matt Palmer, Public Safety Product Manager at NEC Software Solutions, explores the key issues in ensuring that the technology is used ethically and transparently.
Artificial intelligence (AI) is already an integral part of the police toolkit. The police service is increasingly using AI-enabled technology to save time, prioritise resources and increase efficiency.
AI is starting to make a real difference to the police by processing large volumes of data much faster than a human could. This includes some of the essential police functions, such as live classification of incoming calls and automation of data quality assurance work.
We are also seeing more cases where AI is able to support police officers’ decision making by predicting outcomes based on patterns. Examples include the use of supervised machine learning to assess factors such as the likelihood that an individual will offend, reoffend or become vulnerable to victimisation, with these examples proving to be the most controversial in the public mind.
As the law enforcement sector steps up its use of AI, people are increasingly aware of the risks of relying too heavily on technology in decisions that can profoundly affect people’s lives. It is therefore critical to establish approaches in which police use, and are seen to use, AI to the highest ethical standards.
Perhaps one of the most widely expressed concerns about AI in policing is the risk of bias and discrimination. All AI systems learn from the initial training data, and if there is a bias in this data, it will be integrated into the AI models, perpetuating the bias and influencing decision making.
If predictive tools are trained on historical arrest data where human biases exist, the algorithms will replicate discriminatory patterns, such as negative racial profiling and targeting of minority communities.
To prevent bias from infecting AI models, developers must use diverse and representative data sets to train AI systems and continuously test these systems for discriminatory patterns.
Without a doubt, AI will play an increasingly important role in police work. To ensure that this role is applied ethically and responsibly, the final say in any decision must be made by human, not artificial, intelligence.
The problem with AI systems is that they are not infallible. They can produce false positives, such as incorrectly identifying innocent people as suspects. Similarly, they can deliver false negatives and fail to identify the real offenders. Without human judgement, AI could lead to miscarriages of justice.
_____
Aquest apunt en català / Esta entrada en español / Post en français








