Bias in Hiring Algorithms
In 2018, a major tech company scrapped its experimental hiring algorithm after finding it systematically downgraded résumés that included the word “women’s,” such as “women’s chess club captain.” The model had learned from past hiring data dominated by male applicants, and so it treated male-like résumés as the standard. No one had explicitly told the AI to discriminate. It simply absorbed the patterns it saw and applied them.
Bias in Criminal Justice
Another case involved a criminal risk assessment tool used in US courts. Journalists found it overestimated the likelihood of reoffending for Black defendants compared to white defendants with similar records. Judges were relying on these scores when setting bail or sentencing, meaning the bias translated directly into harsher outcomes for certain groups.
Bias in Healthcare
Even in healthcare, bias shows up. An algorithm used by hospitals to identify high-risk patients recommended less care for Black patients than white patients, despite similar health conditions. The model assumed past healthcare spending was a proxy for medical need, but because Black patients historically received less care, the algorithm underestimated their needs.
Why Fairness Matters
These examples show that fairness isn’t an abstract ideal. It’s the difference between systems that help people and systems that quietly replicate the worst of our history.
The Scale of Harm
Fairness is important because unfair AI doesn’t just make mistakes, but it scales them. A single biased decision by a person affects one situation, but a biased algorithm can make that same decision thousands or millions of times, across different places and contexts, before anyone notices. The impact compounds, locking people into cycles of disadvantage. And because AI often feels objective, its decisions can be harder to challenge. Fairness isn’t just about being kind or inclusive. Fairness is about preventing harm from spreading quickly and invisibly through systems that affect everyday life.
