Amazon gender-diversity fail shows limits of tech

Most programmers in Silicon Valley are men, so they may write their own preconceived notions into the code, unwittingly or not

Amazon gender-diversity fail shows limits of tech

By Jennifer Saba

NEW YORK (Reuters Breakingviews) - Amazon is learning that machines are only as good as the people behind them. The US$900-billion e-commerce giant had to shut down a resumé-reviewing computer program early last year after discovering it was prejudiced against women, Reuters reported on Wednesday. Algorithms and artificial intelligence hold promise but are limited by programmers’ own biases.

Jeff Bezos’s company set out four years ago to help streamline the recruitment process by creating machine-based learning to weed through job applications. In theory, it was a great idea, saving time while also selecting the best candidates.

Instead, the program re-enforced regressive traits as its algorithms used data on past candidates for jobs at Amazon to establish patterns. These were heavily skewed to male applicants.

As a result, a resumé detailing the person was a “women’s chess club captain” would be penalized. And any graduates from two all-women’s colleges were dinged, according to the Reuters report.

Amazon, like its other tech peers, has a gender-diversity problem. Men make up at least 60 per cent of the global workforces. Peers Facebook, Apple, Google, and Microsoft have similar stats. Narrow the category down to engineers, computer whizzes and the like and the disparity is even more pronounced.

Amazon had no intention of setting up criteria that sidelined women. Trouble is, most programmers in Silicon Valley are men. So they may well write their own preconceived notions into the code, whether unwittingly or not.

The Bezos behemoth at least deleted the whole idea once realizing it couldn’t fix the flaws. Facebook boss Mark Zuckerberg, on the other hand, insists that algorithms effectively police fake news and other content even though most evidence is to the contrary. And Wall Street is wedded to computer-driven trading models even though they can cause market crashes or make them worse.

Technology is not a quick fix to intractable problems. Amazon’s HR bot only exacerbated the challenge of bringing more women into the company. Changing the status quo is a messy, disruptive process – and one that requires human intervention.

Latest stories