AI industry faces diversity 'crisis,' study says

AI industry faces diversity 'crisis,' study says
© Getty Images

The artificial intelligence (AI) industry is facing a diversity "crisis," and the mostly-white and mostly-male workforce could be creating AI products that perpetuate discrimination, according to a new study published on Tuesday. 

Researchers with New York University's AI Now Institute found that the industry faces a dearth of women, and employs even fewer people of color. Only 15 percent of the AI research staff at Facebook are women, compared to 10 percent of the staff at Google. And only 2.5 percent of Google's workforce is black, compared to 4 percent at Facebook and Microsoft.


"To date, the diversity problems of the AI industry and the issues of bias in the systems it builds have tended to be considered separately," the researchers wrote. "But we suggest that these are two versions of the same problem: issues of discrimination in the workforce and in system building are deeply intertwined."

The study published Tuesday was the result of a year of research by AI Now Institute's Sarah Myers West, Meredith Whittaker and Kate Crawford. 

Their research emerges as the tech industry increasingly contends with the issue of AI's potentially discriminatory effects. In recent years, research has shown that everything from algorithms on Facebook to facial recognition technology produced by Amazon can discriminate against minorities, women and other protected groups.

The researchers with AI Now are arguing that the field must funnel resources into attracting a more diverse workforce, or else they could perpetuate the biases of the mostly white, mostly male researchers behind the sensitive technologies being produced at Google, Microsoft, Facebook and universities around the world.

"Despite many decades of ‘pipeline studies’ that assess the flow of diverse job candidates from school to industry, there has been no substantial progress in diversity in the AI industry,” they wrote.

Listing a few of the examples of potentially biased AI, the researchers named instances in which sentencing algorithms recommended harsher sentences for black defendants, online robots began using "racist and misogynistic language," and Uber's facial recognition technology failed to work for trans drivers. 

They noted that there is currently no data about the gender identity of AI researchers. 

The researchers issued a series of recommendations to contend with the AI field's diversity "crisis," including publicizing more information about AI systems, which are currently notoriously opaque; rigorously testing all AI technology for bias; and even assessing whether some technologies should be produced at all. 

“We see AI systems replicating patterns of race and gender bias that risk deepening and justifying historical inequality,” Whitaker, a co-founder of the AI Now Institute, said in a statement. “The problem of a lack of diversity in tech obviously isn’t new, but it’s reaching a new and urgent inflection point."