September 18, 2017 | Written by: Morgan Childs
Categorized: New Thinking
Share this post:
Seek for reasons to fear the loss of women’s jobs during the artificial intelligence revolution, and ye shall find them. Here’s a headline in Bloomberg on a story about the 2017 World Economic Forum: “The Rise of Robots Will Make the Tech Gender Gap Even Worse.” That meeting took place in the shadow of a sobering report released by the WEF the previous January, summarized in no uncertain terms: “Women in the Firing Line of Fourth Industrial Revolution.” If the headlines aren’t enough to inspire alarm, the report itself certainly is: The WEF predicts that automation will strip jobs from men and women in nearly equal number, but that because women are already underrepresented in technology, the existing gender gap will widen even further. And it isn’t just the WEF that forecasts an unfavorable climate for women’s jobs. As Erika Hayasaki noted in Foreign Policy early this year, the conclusions of a study conducted by McKinsey & Co. foretell the imminent automation of many jobs traditionally held by women.
Automation is likely to shift the jobs landscape radically, but as IBM CEO Ginni Rometty said of AI at the Forum in Davos, “It’s not man or machine.” AI will create many new opportunities for employment, not just take jobs away. But technology’s diversity crisis isn’t an issue for the future: It’ll be a long time before human jobs are entirely automated, but an imbalanced workforce is already taking a toll on the technology and limiting its applications. For better or worse, AI reflects the human biases of its creators, be it by translating a gender-neutral pronoun into “he” for a doctor and “she” for a nurse, or by automatically lightening skin tones in the name of beauty. As anyone who caught wind of engineer James Damore’s 10-page memo to his (now former) colleagues at Google knows, the call to diversity in tech has become a lightning-rod topic, perhaps as revelatory of a troubling work culture as it is of the proclivities of a disgruntled employee. At the U.S.’s biggest tech companies, dominated by white and Asian men, hostility or harassment may also be keeping women and people of color away—or even driving them out—making it all the more likely that AI will represent a limited worldview.
So what can be done to slow or stop the skewed data built into AI that perpetuates all-too-human biases? According to a handful of organizations committed to righting the wrongs of AI created exclusively by “guys with hoodies,” the key is not only to remove the barriers to hiring and retaining skilled employees from all backgrounds, but also to establish new channels for education within communities and demographic sectors that have traditionally been underrepresented in the field. That means opening new opportunities for people of all genders, races, sexual orientations, socioeconomic backgrounds—and ages.
“A lot of times somebody’s career path starts when they’re in high school,” says Tess Posner, Executive Director of AI4ALL, an initiative based out of San Francisco that officially launched in March of this year. “For example, by the time you get into the job market it [can be] so competitive that if you don’t have a computer science degree, or you don’t have a lot of experience, it can be harder to compete.” In other words, if certain groups of people fall through the cracks, it can be a challenge to climb up the ladder later in life. And that leads to a lopsided landscape for people working in AI—and for those interacting with it. “We really believe that the majority of people are going to be consumers of AI, which means that AI’s getting consumed by a very diverse population and yet it’s being built by a very homogenous population,” Posner says of her colleagues at AI4ALL. “Which really means the technology won’t be able to meet the diverse needs of the society.”
To combat AI’s skewed demographics, AI4ALL is playing the long game. The organization works with leading universities to establish education programs for high schoolers, long before they miss out on any opportunities that would allow them to work in the field in the future. Posner points to a study conducted by Microsoft of 11,500 teenage girls across 12 European countries revealing that many girls on the continent develop an interest in STEM subjects (science, technology, engineering, and math) around age 11—and then begin to lose that interest by age 15. The girls cited a lack of female role models as a primary factor in their disinterest, and Microsoft concluded that the young women lacked sufficient “practical, hands-on experience with STEM subjects.” Posner says that AI4ALL aims to catch young people in that window, which is why the program currently targets rising tenth graders. And it’s not just girls—AI4ALL seeks to support all gender, racial, and socioeconomic populations that are underrepresented in the field.
The organization is also working to give AI a bit of a rebrand. “When we collectively picture AI as one type of thing—whether it’s humanoid robots or self-driving cars or deep learning—we’re encouraging the next generation of researchers to be excited exclusively about those narrow things,” AI4ALL founder Olga Russakovsky wrote in a piece for MIT Technology Review last August. AI4ALL shifts focus away from the dystopic techno-futurism pervasive in pop culture and redirects it toward what the organization calls “humanistic AI.” Participants take part in projects that show AI’s human, real-world impact, such as assisting in cancer diagnoses or creating greater mobility for aging populations, Posner says.
AI4ALL isn’t the only organization working with the next generation to address the issue of a growing diversity crisis in AI. IBM has partnered with Girls Who Code, a nonprofit targeting young women between the sixth and twelfth grades, and pioneered P-TECH, an early-college program providing high schoolers with instruction in flexible skills for jobs in technology, including a six-year “workplace learning” sequence. “New-collar” jobs, a term coined by IBM’s Rometty, refers to a workforce where skills, rather than degrees, open opportunities for employment, and the company has made strides to temper the anxiety surrounding the rise of AI, such as in a letter to congress this summer urging lawmakers to look optimistically towards the future of the technology. A longtime proponent of inclusive workplaces, IBM seeks to set a precedent for prioritizing diversity in the hiring process that other companies will follow, according to Vice President of Human Resources Sam Ladah.
Posner sees AI4ALL and similar efforts to educate the next generation of tech employees as a single component of a multi-pronged approach to balancing out industry demographics. She calls for companies to have “clear metrics…both on the hiring side and on the retention side,” and also to create mentorship opportunities to boost employee confidence and skill development. She also stresses the need to continue research into the consequences of limited diversity in tech. “I think it’s really important to get that out there, and think critically about how we can prevent the technology from really mirroring some of the things that we want to change about our society.”
As for AI4ALL’s young alumni, Posner says that a developing mentorship program will serve to support them well into the future, too. It’s hard to predict the myriad ways in which AI could reorganize the jobs landscape, but Posner isn’t concerned that the organization’s efforts will be lost on their alumni. “We know they’re going to accomplish incredible things whether they go into AI or not,” she says, “in the different ways that this technology will touch our society in the future.”