A University of Chicago research team has built a machine learning system that predicts crime with roughly 90% accuracy — and in doing so, has illuminated something older and more troubling than any algorithm: the uneven hand with which law enforcement has long touched rich and poor neighborhoods alike. Published in July 2022, the work reveals that across eight major American cities, crime in wealthy areas reliably produces arrests, while crime in poor areas often does not. The researchers have chosen to make their tool publicly auditable, framing it not as an instrument of control but as a mi
Los Angeles Times
· Jul 04, 2022
· technology
· 4 min read
High
Sourcing: Verified
AI tool aims to predict crime and expose policing bias in major cities
Systemic underpolicing in disadvantaged neighborhoods leaves residents with inferior police protection and unequal access to law enforcement resources.
◉
Politics & Governance
10995 related articles
YouTube educator Ms. Rachel's donation to progressive Michigan Senate …
Fox News · 100% match
Burnham Vows Power Shift North as UK's Likely Next PM
The New York Times · 99% match
BJP sweeps Gujarat for seventh time; Congress retains Himachal Pradesh
The Indian Express · 99% match
Burnham urged to call election; heatwave dominates UK front pages
BBC News · 99% match
Xi's diplomatic offensive: China courts 'middle powers' as alternative…
The Guardian · 99% match
Related Coverage
The Guardian · Jul 27
Court upholds finding that Hanson's tweet to Faruqi breached racial discrimination law
A federal court unanimously upheld a finding that One Nation leader Pauline Hanson engaged in racial discrimination agai…
The Guardian · Jul 27 Cockroach Protests Expose Cracks in Modi's Grip on IndiaMass protests in India forced education minister's resignation, marking rare public victory against Modi's BJP and expos…
Al Jazeera · Jul 27 Kashmir's Refugee Seat Dispute Threatens Pakistan Election LegitimacyPakistan-administered Kashmir holds elections amid violent protests over 12 assembly seats reserved for refugees, raisin…
The Guardian · Jul 27 Ben Carroll emerges as Jacinta Allan's sole challenger for Victoria's top jobBen Carroll, Victoria's deputy premier and education minister, has positioned himself as a potential challenger to Premi…
Bias & Framing
⚖
Article presents AI crime prediction tool optimistically as bias-detection mechanism while emphasizing historical concerns about algorithmic bias in policing, with framing that favors the researchers' equity-focused intentions.
Bias Direction
center-left
Confidence
MED
Solution-oriented framing that positions the research as corrective to past algorithmic harms. Opens with 'For once' to contrast this tool against previous crime prediction systems, establishing a redemptive narrative. Emphasizes researchers' stated intentions to prevent misuse and promote equity.
For once, algorithms that predict crime might be used to uncover bias in policing, instead of reinforcing it
inferior police protection in poorer neighborhoods
biased data that contribute to increased patrols in Black and Latino neighborhoods
false accusations against people of color
systemic biases in law enforcement
Law enforcement agencies' perspective on the tool's practical applicability
Skepticism from civil liberties advocates about whether audit capabilities prevent actual misuse
Technical limitations or potential failure modes of the 90% accuracy claim
Perspectives from communities in the studied neighborhoods
Counterarguments about whether prediction accuracy alone addresses enforcement disparities
Geopolitical Impact
🌎
University of Chicago researchers developed an AI crime prediction tool revealing systemic policing bias in eight U.S. cities, potentially shifting domestic policy toward equitable resource allocation rather than enforcement escalation.
Escalation Risk
low
Confidence
MED
United States
Los Angeles
Eight major U.S. cities
This represents a subtle shift in power dynamics within the U.S. domestic sphere: from law enforcement agencies wielding unchecked algorithmic surveillance tools to academic institutions and civil society gaining capacity to audit and expose enforcement bias. The technology democratizes accountability mechanisms, potentially constraining state power through transparency rather than expanding it through prediction.
Similar to 1960s-70s civil rights era when data-driven studies (e.g., Kerner Commission) exposed systemic racial disparities in policing, prompting policy debates—though implementation remained inconsistent.
Economic Lens
📈
AI crime prediction tool reveals policing bias in major cities, potentially redirecting law enforcement resources toward equitable, need-based allocation rather than reinforcing discriminatory practices.
Market Signal
neutral
Confidence
MED
Law enforcement and public safety
Software/AI technology
Municipal government and public administration
Social services
Criminal justice system
Households in underserved neighborhoods may experience improved police response and resource allocation; reduced discriminatory policing could lower legal costs and incarceration rates for affected communities; potential reallocation of public funds toward alternative responders (social workers, mental health professionals) may improve service quality.
Likely to prompt municipal governments to audit existing predictive policing systems for bias; potential regulatory frameworks requiring algorithmic transparency in law enforcement; possible shifts toward alternative emergency responders; may influence federal funding criteria for police departments; could drive policy changes in resource allocation methodology across major cities.