To Predict and Serve?

In the citing paper

“by using arrests as a proxy for crimes committed, COMPAS can only exacerbate racist policing practices, rather than transcending them [7,13,23,37,39].”

In the citing paper

“If an AI system is trained on such a biased sample, it will learn that people with an immigrant background are more likely to commit crime.”

In the citing paper

“Lum and Isaac (2016) showed that PredPol, a popular predictive policing software, reinforced existing biases in the dataset and targeted Black people at disproportionately high rates (twice as high as white people, even though drug use was roughly equal between the two groups)”

In the citing paper

“Such interactions between human and machine biases may have compounding effects - for better or worse - in the presence of human-AI feedback loops in learning or decision-making”

In the citing paper

“The effects of such a biased sample could even be amplified by AI predictions.”

In the citing paper

“if biased data is used to train these predictive models, the models will reproduce (…) those same biases.”

In the citing paper

“If police focus attention on certain ethnic groups and certain neighbourhoods, it is likely that police records will systematically over-represent those groups and neighbourhoods.”

In the citing paper

“The sampling procedure can also be biased.”

In the citing paper

“A predictive-policing application learns from past data that arrests have occurred frequently in certain areas, and it proposes that police patrol these areas preferentially. This leads to more arrests in these areas, which in turn feeds the learning to propose to patrol them, etc.”

In the citing paper

“they further showed via simulation that if police are more likely to find crime in areas where they are actively policing (as a result of PredPol's predictions), which in turn affects the crime rates in the data used by the algorithm, then the algorithm will become further biased in a”

In the citing paper

“Similarly, historical over-policing of some communities has unavoidable impacts on the sampling of the data used to train models for predictive policing [32,79].”

In the citing paper

“When discrimination affects arrest rates, it becomes embedded in prosecution data (Lum and Isaac 2016).”

In the citing paper

“Similar audits have turned up issues of bias reinforcement in systems that guide police to portions of a city to patrol [36], and in e-commerce platforms [35].”

In the citing paper

“For example, arrests are often used as a proxy for crime in applications of machine learning to policing and criminal justice, even though arrests are a racially biased representation of the true incidence of crime [28].”

In the citing paper

“While the concern for justice has a long history in the West (e.g., Aristotle, Rawls) and the East (e.g., Confucius, Chanakya), they show that the majority of empirical work on social justice has been situated in the US and Western Europe, grounding the understanding of justice in the Western cultural context.”

In the citing paper

“Summarising decades worth of research, [120] says that the more collectivist and hierarchical societies in the East differs from the more individualistic and egalitarian cultures of the West in how different forms of justicedistributive, procedural, and retributive-are conceptualised and achieved.”

In the citing paper

“Algorithms that aim to facilitate hiring, lending, healthcare, policing, and criminal justice may provide an illusion of objectivity while reinforcing human biases and creating feedback loops that further exacerbate injustice and inequality”

In the citing paper

“Despite the opacity, research has revealed how algorithms lead users to radical or age-inappropriate content (128, 164), exacerbate disparities in health (123), and increase bias in policing (124).”