Further automation through so-called predictive policing systems have their own pitfalls.

通过所谓的预测性警务系统进一步实现自动化有其自身的缺陷。

Law enforcement is also starting to use machine learning for predictive policing.

执法部门也开始将机器学习用于预测性警务。

Meanwhile, in the social domain, AI-driven predictive policing or facial recognition technologies are reshaping societal norms and expectations around privacy and surveillance.

与此同时，在社交领域，人工智能驱动的预测性警务或面部识别技术正在重塑围绕隐私和监控的社会规范和期望。

They will reproduce the biases and prejudices of those who create them, like the webcams that only recognise white faces, or the predictive policing systems that lay siege to low-income neighbourhoods.

它们将复制那些创造它们的人的偏见和偏好，就像只能识别白人面孔的网络摄像头或对低收入社区实施预测性执法的系统一样。

In places like the US, where African-Americans are often disproportionately and, in some cases, lethally targeted by the police, predictive policing could interpret existing data, to potentially perpetuate those discriminatory patterns.

在像美国这样的地方，非裔美国人往往更有可能成为警察的目标，在某些情况下甚至是致命的目标，预测性警务可以解读存量数据，从而有可能使这些歧视性模式永续下去。

This year, the AI Now Institute in New York published a report that revealed that the AI technologies that are being used for predictive policing have been trained on " dirty" data.

今年，位于纽约的人工智能现在研究所发表的一份报告揭示了预测警务领域的人工智能技术是使用非常糟糕的数据进行训练的。