Last year, the Equal Employment Opportunity Commission (EEOC) expressed its concern that the growing use of artificial intelligence (AI) in recruiting software could unintentionally discriminate against persons in a protected category. Many AI programs learn over time by incorporating information gathered from the internet and other sources. If those sources contain biased information, that could be reflected in the decisions made by the recruiting program.
On Monday, the EEOC announced settlement of multiple discrimination charges filed against a Florida company that operates a job search website for technology professionals. The charges allege that keywords provided by clients for their job searches include terms that deter persons of American national origin from responding to the postings. In the settlement agreement, the company agreed to “scrape” its programming to remove discriminatory keywords that could be used to exclude persons based on national origin (i.e., references to visa status).
Employers that license recruiting software may not have the right or the capability to review or make changes to the programs to avoid discriminatory results. Even if they do, it can be difficult for even experienced programmers to determine exactly how some AI programs produce their results, making detection of discriminatory impact more difficult. In order to reduce liability, employers using these recruiting programs should carefully review the terms of sale or license to make sure (1) they include a representation by the vendor that the program has been validated for use in the hiring process; and (2) that the vendor accepts financial responsibility for failure to adhere to that representation.