Algorithmic Bias
Algorithmic bias is when computer systems produce unfair, discriminatory outcomes due to skewed data or flawed design, often harming marginalized groups like Black women in facial recognition systems.
Algorithmic bias refers to systematic and repeatable errors in a computer system that create unfair outcomes, such as privileging one arbitrary group of users over others. This bias typically arises from skewed training data, flawed model design, or the uncritical translation of human prejudices into code, leading to discriminatory results in areas like facial recognition, hiring, and lending.
What this means in real life
A resume-screening algorithm trained on past hiring decisions may learn to downrank female applicants if the company historically hired fewer women—perpetuating that pattern automatically, even if no one programmed it to discriminate.
What it isn’t
Not simply 'the algorithm made a mistake.' Algorithmic bias is systematic unfairness that consistently harms specific groups, not random errors affecting everyone equally. A one-off wrong prediction is a mistake; a pattern favoring one demographic is bias.
Commonly misused online
Often used as a catch-all for 'any bad AI outcome' or 'AI I disagree with,' when it specifically means systematic disadvantage to protected groups. People sometimes invoke it to mean 'the algorithm is wrong about me personally,' which conflates individual error with structural discrimination.
Based on 1 reference source, including government sources. Last verified July 12, 2026.