News
Computing Applications News

Bias in Technology

As leading companies release troubling diversity statistics, experts search for solutions.
Posted
  1. Introduction
  2. Diversity Is Good Business
  3. The Pipeline Problem
  4. If You See It, You Can Be It
  5. Author
  6. Figures
Bias in Technology, illustrative photo
Organizations like Code.org are working to expand access to computer science and increase participation by women and underrepresented students of color.

The technology world has a diversity problem. A recent U.S. Equal Employment Opportunity Commission (EEOC) report found that the high-tech industry employed far fewer African-Americans, Hispanics, and women, relative to whites, Asian-Americans, and men. The difference is especially glaring in Silicon Valley. At Google and Facebook, African-Americans represent just 1% of the tech work force. The numbers are slightly higher at some other leading technology firms, but still are hardly reflective of society at large.

In academia, the figures are also discouraging. According to the 2015 Taulbee Survey, conducted by the Computing Research Association, African-Americans represented only 4.6% of the students awarded bachelor’s degrees in computer science (CS). Women represented 15.7% of the surveyed population, but this is a significant decrease from 1984–1985, when the National Center for Education Statistics found that women made up 37% of CS undergraduates.

Today, experts in both the corporate and academic worlds are working to understand the root of the imbalance and searching for ways to expand the number of women and minorities in technology—and keep them there.

Back to Top

Diversity Is Good Business

The necessity of a more diverse technology workforce is a matter of social equality, but there are other compelling factors as well. When groups are under-represented on product development teams, for example, the resulting technology can be biased. “I always ask the question: Is it possible for you to invent or create anything in the absence of who you are?” asks University of Florida computer scientist Juan Gilbert. “You might say, ‘I write algorithms; there’s no bias in that.’ But we’ve seen examples where algorithms do have bias.”

In 2015, due to a quickly corrected flaw in its facial recognition software, Google Photos sorted images of one user and her African-American friend into an album labeled “Gorillas.” In another 2015 incident, a white male and his African-American friend recorded video of themselves testing the electronic soap dispensers at an Atlanta hotel because the devices did not register the presence of black skin.

Gilbert and others argue that a more diverse group of employees on the product development side could have prevented these kinds of problems. But there are also indications that broader representation can lead to increased revenue. Gilbert cites the story of the once-popular Motorola Razr mobile phone. When the product team was brainstorming potential colors for the phone, one of the women in the group suggested a more feminine hue for one of the models. She was laughed at initially, Gilbert says, but eventually the men in the group conceded. “They created a pink phone and it became the number one seller,” Gilbert says. “If she hadn’t been in the room, that wouldn’t have happened.”

Still, the business argument for inclusion may be simpler than that. “If you really want everyone in the world to use your products, then you need to have everyone working for you,” says Kaya Thomas, an African-American woman majoring in computer science at Dartmouth College. “If you want to sell to everybody, you have to hire everybody.”

Back to Top

The Pipeline Problem

Last July, when Facebook revealed it had made little to no improvement in its diversity numbers relative to the prior year, the company attributed the shortfall in part to a pipeline problem. The statement prompted significant criticism, with many citing Taulbee Survey data as evidence that qualified students are being overlooked: if 4.6% of CS graduates are African-American, why are only 1% or 2% of new tech-related hires at some major companies African-American? “The students are out there,” says Ann Quiroz Gates, chair of the Department of Computer Science at the University of Texas at El Paso. “I don’t agree the talent is not there. The talent is there.”

Gates says many of the skills companies look for in students can be acquired outside the classroom by competing in hackathons or building a GitHub profile. Unfortunately, her minority students often do not have the spare hours to participate in such extracurricular activities. “Finding the time to program beyond classroom assignments and experimenting with computers is really difficult because many of our students work full time while carrying a full course load,” she says. “They’re exhausted.”

The high-tech companies Gates has engaged with are starting to understand that, and she is working with several companies to set up paid positions at local firms, or university-funded projects that would allow her students to acquire those outside-the-classroom problem-solving skills and still earn the money they need to pay for their education.

While these efforts may help existing students land sought-after positions, the Taulbee Survey indicates more needs to be done to encourage women and minorities to pursue CS in the first place—and to stick with the subject. Maja Mataric, a computer scientist at the University of Southern California, began her career in the 1980s, when the participation of women in CS was on the upswing, so she has observed the decline firsthand. “We did really well for a while, but we never reached equality at all,” she notes.

In her view, several factors influenced the drop. The culture of male-dominated departments, for example, either ended up pushing women away or encouraging them to assimilate and sustain the very culture that was discouraging others. Mataric also points to how departments often promote women-in-computing organizations such as the Anita Borg Institute as support groups. This suggests the women selected to attend need this support, and are somehow weak. Instead, Mataric says, their selection should be presented as a sign of excellence.

These subtle cues add up, according to Mataric. “It is a very, very slanted, un-level playing field,” she says, “and the only people who can survive are either so driven that they don’t care or they’re just bullheaded. But that’s a small subset.”

Changing the academic culture can only solve part of the problem. Mataric believes the effort to increase the representation of women in technology needs to begin at an early age. In her own outreach projects, she focuses on introducing elementary school students to computational thinking through robotics. Part of her strategy is to show kids that software engineering is not just coding in the dark. “I really think the only way to fight it is from the very beginning, because so much of what we’re doing is just patching along the way.”

Kinnis Gosha, an assistant professor of computer science at Morehouse College, a private, all-male, liberal arts, historically black college in Atlanta, lauds projects such as Black Girls Code for spiking interest in the field among middle school students, but he says such programs are just the first step. Gosha, who is also the director of the college’s Culturally Relevant Computing Lab, calls for increased focus on improving high school-level computer science education. If college students want to compete for prestigious internships at the top technology companies and then land the best jobs, he argues, they need to start building their skills in high school by taking classes such as Advanced Placement (A.P.) Computer Science.

In his own courses at Morehouse, Gosha has seen firsthand that a student with this experience has a tremendous head start. “A student able to take that class and do well is a year or more ahead of the others,” he says. Unfortunately, the program is not always an option for kids. “In the Atlanta public school system,” Gosha notes, “there are only two high schools that offer the A.P. Computer Science course.”

Back to Top

If You See It, You Can Be It

Still, a student who starts late can enjoy a bright future. Kaya Thomas, the Dartmouth undergraduate, was not interested in computer science in high school. Then, during the winter break of her freshman year, Thomas started reading about the lack of women, and particularly women of color, in computer science. She took an online coding course during the break, fell in love with the work, and switched her major when she returned to school. Now, as a Fellow of CODE2040, the nonprofit that helps African-American and Latino computer science majors, she’s part of a community of woman and minorities like her, and she has already interned at three major technology companies.

She is the only African-American woman in her class of approximately 90 computer science majors at Dartmouth, but she is mentoring younger students, and encouraging them to stay in the field. “There are a lot of black girls—and by a lot I mean five or six—majoring in computer science in the sophomore class,” she says with pride. But such small examples of success could continue to build on one another. “If you never see anyone who looks like you in a space, it’s a challenge for you to be in that space,” says Gilbert. “If you see it, you can be it.”

Of course, successful minority or female software engineers and developers cannot spend all their time mentoring. With that in mind, Gosha and several colleagues have been working on a virtual mentoring program that would feature interactive avatars of minority tech workers and executives responding to common questions. The avatars would not replace face-to-face mentoring; they would complement it, or even offer a young person their first exposure to a successful minority tech worker. “The idea is to virtualize the experience of meeting another black person who actually has a computing job, because African-American youths might not be in places where those people are visible,” he explains.

Overall, experts and insiders believe improving diversity in the technology world will require continued effort from academia, corporations, nonprofits, universities, and others, but Gilbert says the impact of these projects will reach beyond the business world. “We can’t be as great as we can be as a country unless we increase the participation of everyone,” he says. “Diverse minds will give us the best solutions to problems.”

*  Further Reading

Diversity in High Tech, U.S. Equal Employment Opportunity Commission, 2015. http://www.eeoc.gov/eeoc/statistics/reports/hightech

Hewlett, S.A., Luce, C.B., Servon, L.J., et. al.
The Athena Factor: Reversing the Brain Drain in Science, Engineering, and Technology, a Harvard Business Review Research Report, 2008.

Intel Corporation & Dalberg Global Investment Advisors Decoding Diversity: The Financial and Economic Returns to Diversity in Tech, 2016. http://bit.ly/2bHEFuN

Zweben, S. and Bizot, B.
2015 Taulbee Survey: Continued Booming Undergraduate CS Enrollment; Doctoral Degree Production Dips Slightly, Computing Research News, May 2016, Vol. 28 / No. 5

Back to Top

Back to Top

Figures

UF1 Figure. Organizations like Code.org are working to expand access to computer science and increase participation by women and underrepresented students of color.

Back to top

Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More