Amazon stops AI program evidencing gender-diversity fails

A résumé-reviewing program from Amazon was discovered to have prejudiced against women.

Amazon shut down a computer algorithm used to recruit new staff because it developed a bias against women.

The $900 billion e-commerce behemoth set out four years ago to create an Artificial Intelligence (AI) with machine-based learning to scroll through job applications and help speed the process of recruitment.

The AI program would rate the candidates out of five stars, similar to how products are ranked on Amazon’s retail site, unfortunately, it was re-enforced with regressive traits and with programmers’ own biases, as its algorithms used data on former candidates for Amazon jobs in order to establish future partners for the company.

Results from the investigation were clear: during the past 10 years, recruitment has been heavily skewed to male applicants. It was finally shut down in 2017.

Quite differently than Facebook, Amazon at least deleted the whole idea once realizing it couldn’t fix the flaws. The Mark Zuckerberg company, on the other hand, insists that algorithms effectively police fake news and other content even though most evidence is to the contrary.

The failed program, according to related sources, was created literally as company leaders wanted it to be an engine that gave you 100 resumes, spitting out the top five, the ones hired.

After calling out the AI technology, Amazon worked on changing the algorithms to ensure that gender-related terms were ranked neutrally, but no great upgrade was achieved, causing the final shutdown.

An industry not evolving as it should

Amazon and its tech peers have a great diversity problem, not that other industries don’t.

Amazon’s HR bot only exacerbated the challenge of bringing more women into the company, but the truth is even more harsh, as 60% of the global workforces are represented by men, with similar stats in companies like Facebook, Apple, Microsoft and Google.

True: Amazon had no intention of setting up criteria that sidelined women, but this only just came to prove the challenges women have to face in one industry, just imagining opening Pandora’s box on other places.

Maybe it’s time to reevaluate how important human intervention can be in these and other matters.