Amazon has reportedly had to scrap an artificially intelligent recruitment tool built by its in-house engineering team, after sources told Reuters the results began to be skewed in favour of male candidates.
Five members of the team who worked on the tool in 2014 said it began to teach itself that male candidates were preferable over women, due its reliance on internal historical data of which employees Amazon might prefer.
"They literally wanted it to be an engine where I'm going to give you 100 resumes, it will spit out the top five, and we'll hire those," said one of the engineers.
The tool's job was to review incoming job applications and rate them out of five stars. The engineers told Reuters the idea was scrapped last year once it was determined the gender bias could not be corrected.
Amazon declined to comment.
It is unclear whether Amazon actually ever used the tool in a real-life hiring process, however one source said the company now uses a "much watered-down version" of the engine to help with rudimentary tasks such as removing duplicate applications.
Amazon's current global workforce is a 60:40 split between men and women, according to company data.