Tag: bias

  • The Real-Life Risks Of Predictive Policing

    Imagine being targeted not for a crime committed but for a prediction. Predictive policing uses AI to flag people and places but has sparked backlash over bias and lack of oversight. While many cities fell short San Jose sets a new example by making fairness transparency and public accountability central to its approach.

  • New AI Benchmarks Aim to Reduce Bias

    Stanford’s new AI benchmarks aim to tackle bias by introducing two types: descriptive and normative. Descriptive benchmarks measure factual accuracy, while normative ones assess harmful stereotypes. These benchmarks challenge traditional fairness methods, urging AI systems to consider group differences and context for better, more accurate outcomes.