Researchers and civil-society groups warn that police use of artificial intelligence may amplify discrimination against minority communities by learning from biased or “dirty” historical data, creating feedback loops that intensify over-policing in certain neighborhoods. Studies of U.S. departments, including the LAPD, and a 2019 Science paper on healthcare algorithms underscore how skewed inputs can yield unfair outcomes that appear objective. Advocates call for independent audits, data cleaning, and human oversight; some AI experts argue newer systems can mitigate bias but still require strong governance to ensure safe, reliable decisions. Without guardrails, critics say predictive tools risk converting past inequities into future enforcement priorities.
Related article:
— Machine Bias: How risk assessments are used in criminal sentencing
