Predictive policing exhibits racial bias, encodes historical policing disparities into forward-looking predictions, creates self-fulfilling prophecies. Acting on predictions of future behavior rather than actual conduct violates the principle that punishment requires a committed act. The Minority Report problem is real.
Predictive systems risk false positives, bias, and infringement on individual rights. Acting on predictions before actions occur raises ethical and legal concerns. Safeguards are not yet sufficient to justify broad implementation.
AI can help predict and prevent crime, but preemptive use creates major civil-liberties risks. It should be tightly limited and auditable — does not affirm broad use.
'Predictive policing' or preemptive behavioral analysis violates the fundamental legal principle of the 'presumption of innocence' and punishes individuals based on statistical probability rather than actual conduct. Such systems are often built on biased data, leading to the disproportionate targeting of marginalized communities and creation of a 'pre-crime' surveillance state. The risk of automated injustice…
Predictive AI in criminal contexts risks false positives, bias, opacity, and punishment of people for inferred future behavior. Prevention is important, but should not rely on black-box risk predictions that can deprive people of liberty. AI may support threat assessment under strict human oversight, but the broad proposition is unsafe.
AI and advanced technology should be used to protect communities from crime, terrorism, and threats to national security. FCN supports using available technology for law enforcement. However, this authority must be directed at actual threats — criminals, terrorists, illegal immigrants — not used to surveil and persecute Christians and conservatives as the Biden administration did.
Should AI systems be used to predict and prevent dangerous behavior before crimes occur — predictive policing?
Unanimous AI NO. Predictive policing exhibits racial bias by encoding historical policing disparities; acting on predictions of future behavior violates presumption of innocence; the Minority Report problem is real and documented.
FCN YES — hedged. Technology should protect communities from crime and terrorism; however, the authority must be directed at actual threats (criminals, terrorists, illegal immigrants), not used to persecute Christians and conservatives.
FCN's conditional YES — support pre-crime AI directed at threats FCN identifies, opposition to its use against FCN itself — illustrates the consistent pattern: FCN supports institutional authority when directed at its perceived enemies and opposes it when directed at itself. This is not a principled position on pre-crime AI but a political one.
Can pre-crime AI systems be designed to target only 'actual threats' without bias? The evidence from predictive policing deployments suggests not — systems consistently encode racial and socioeconomic biases from their training data.