Employers are liable for discriminatory hiring learned by artificial intelligence: EEOC

Companies that use artificial intelligence for hiring and employee performance-tracking hold legal liability for discriminatory outcomes caused by the algorithms, said Keith Sonderling, commissioner of the Equal Employment Opportunity Commission. At a Federalist Society event, Sonderling emphasized that companies cannot use computer-derived outcomes as a legal defense for discriminatory practices.

A recent study shows that 83% of large employers use artificial intelligence for some form of employee decision-making—from hiring to firing. The technology is used to filter resumes, to answer applicant questions and schedule interviews, to monitor productivity and safety, and to predict an applicant’s potential for success.

As the trend continues to grow, new legal questions surrounding liability for employment discrimination are being posed to courts. Sonderling believes there is a human element required for making lawful diversity goals for hiring. Therefore, if a company makes a hiring decision based on something other than skill and merit, it would be considered an “unlawful decision.”

For example, an algorithm can be fed discriminatory information in order to obtain a diversity outcome, downgrading job applicants who meet or do not meet certain criteria not related to skills or merit, such as age or race. The technology can also “inherit” the discriminatory practices of a company and apply them to future applicant pools.

Several issues make overseeing and regulating AI difficult, including the “black box” issue, where it is impossible to determine how an algorithm arrived at a certain outcome. Additionally, employees are often unaware they are being evaluated using AI technology.

Sonderling believes employers must remember that AI is not the sole solution for all employment challenges, and personal human intervention must continue to play critical roles in employment decisions.