Amazon Abandons Biased AI Hiring Tool That Favored Men Over Women
Amazon scraps an AI recruitment system after discovering it learns to discriminate against female candidates. The incident highlights how biased training data creates flawed algorithms.
Amazon abandons an AI recruitment system after discovering the software actively discriminates against female candidates. The company develops the automated tool in 2014 to rate job applicants from one to five stars, but ultimately scraps the project last year when engineers realize it penalizes women.
The bias stems directly from the data used to train the algorithm. Because Amazon feeds the system 10 years of historical resumes, the AI learns to favor male candidates since men dominate the tech industry's applicant pool. The program specifically downgrades resumes that include the word "women's" or mention all-women's colleges.
This situation demonstrates that artificial intelligence is not inherently unbiased. When companies train these systems on flawed historical data, the algorithms simply learn and amplify existing societal prejudices. Without careful oversight, biased AI tools threaten to worsen workplace diversity issues rather than improve hiring practices.