Inside the Gathering Spot, a posh members-only club in Atlanta that serves a diverse set of entrepreneurs and innovators, Travis and Troy Nunnally—better known as the Tech Twins—are holding court.
After discovering a love of engineering by building soapbox derby racing cars as kids, the brothers have launched a few different companies in Atlanta. Holding two master’s and a doctorate from Georgia Tech between them, the twins are cofounders of Brain Rain Solutions, which builds augmented-reality, internet-of-things, and technology-based products for companies. Their most recent product, FaceMD+, uses dermatologist databases and machine learning to create a customized skin-care tracker—with algorithms providing customized data for each skin tone and type.
The Nunnally brothers specialize in applying machine learning to a variety of problems, and their growing business means that they spend a lot of time hiring employees and contractors.
But as two black men on the forefront of the machine-learning revolution, they are also concerned about the readiness of black tech talent: Machine learning requires knowledge of Python coding, algorithmic optimization techniques, and advanced math like calculus. “The first step into the pipeline is a developer,” Troy says. “Once you have that base, you can add on the skill set. In the African American community, [the funnel] gets narrow and even more narrow.”
The Nunnallys represent a rarity in tech, much less AI. “When I graduated in 2014, there were less than 100 black men in the whole nation with a PhD in machine learning,” Troy says. “We get scared because we don’t see anybody like us. We do see people at the top in entertainment, in sports, but we don’t see people at the top of technology.”
What would it take to make more tech twins? And how can technology become more diverse and welcoming to underrepresented groups? The answer may be where few are looking—the city of Atlanta.
Some 2,482 miles outside of Silicon Valley, Atlanta is a technological powerhouse—with a growing focus on the burgeoning field of artificial intelligence. Blessed with excellent institutions of higher learning like Georgia Tech, Emory, Morehouse, and Spelman, and with a robust private sector, Atlanta’s tech scene is also a rarity: a hotbed of diverse innovation.
Coca-Cola, Home Depot, and UPS are all headquartered in Atlanta, and tech standouts like Jewel Burks Solomon (who sold her startup PartPic to Amazon) and Tristan Walker (who is relocating his Walker Brands HQ to Atlanta after his sale to Proctor & Gamble) call the city home.
But when it comes to AI, even Atlanta’s tech scene is not making much of a mark in the national picture of gender and racial diversity. According to the AI Now Institute’s 2019 report Discriminating Systems: Gender, Race, and Power in AI:
Recent studies found only 18 percent of authors at leading AI conferences are women and more than 80 percent of AI professors are men. This disparity is extreme in the AI industry: Women comprise only 15 percent of AI research staff at Facebook and 10 percent at Google. There is no public data on trans workers or other gender minorities. For black workers, the picture is even worse. For example, only 2.5 percent of Google’s workforce is black, while Facebook and Microsoft are each at 4 percent. Given decades of concern and investment to redress this imbalance, the current state of the field is alarming.
What explains this sorry state of affairs?
Pointing to a so-called pipeline problem—the well-trod argument that tech companies can’t find qualified women and people of color because they aren’t coming out of the training pipeline—is a convenient excuse for firms reluctant to tackle the issue.
Not to say that there aren’t troubling reasons for the shortage of trained engineers of color. African American students are actively discouraged from taking more challenging AP classes and often lack either resources for tutoring or access to specialty prep classes. Often, advanced-level math and science classes are not offered at the high school level, leaving a much harder climb in college.
As a grad student at Georgia Tech, Betsy DiSalvo noticed African American males dropping out at high rates from undergraduate Introduction to Computer Science classes. She wondered why strong interest in other realms of technology, like playing videogames, didn’t translate to success in computer science. Her question evolved into a dissertation project, Glitch Game Testers, a three-year partnership between Georgia Tech and Morehouse to encourage more African American high school upperclassmen to enter computer science as undergrads by working as paid videogame testers and taking workshops in computer science.
Now an associate professor at Georgia Tech's School of Interactive Computing and the founder of the Culture and Technology Lab, DiSalvo is one of the few researchers to deeply examine the role of social norms and videogame play practices to understand why certain demographic groups may feel included or excluded from computing as a profession.
Cultural myth suggests that black teens tend to view academic achievement as being coded “white.” DiSalvo and her team questioned that idea, finding it simplistic and lacking in cultural context. In a 2014 paper called “Saving Face While Geeking Out: Video Game Testing as a Justification for Learning Computer Science,” DiSalvo and her coauthors examined student behavior and attitudes toward computer science and found complex motivations for smart students who seemingly distanced themselves from their intellectual and professional goals:
Participants repeatedly told us that they tried not to talk about the CS aspect of the program with their family. Most suggested that it was because their families would ask too many questions, but more subtle fears of knowing more than one’s parents or having parents look foolish were also put forth as reasons for avoiding these conversations.
The takeaway is that culture (and knowledge about culture) is key to understanding the issues of belonging that face African Americans entering advanced technology fields. The work of DiSalvo and her team has implications across the broader computing field, especially when it comes to issues of retention on the path to an advanced degree.
Since DiSalvo’s Glitch Game Testers project was built around a specific hypothesis involving videogame play and computer science training, it is difficult to expand its findings to other fields or even to other communities of African American boys. That said, the program boasted impressive returns, with more than half of its 32 participants continuing on paths to computing careers into postsecondary education. But the small sample size means a lot more research is needed into the motivations of African American males and identification with computer science.
After completing Glitch and launching additional research projects to address gender-, race-, and class-based disparities in computer science exposure, DiSalvo says she realized that a lot of the diversity and pipeline issues are more social than technical in nature. Students who felt like they didn’t belong or who didn’t fit the environment fostered in the classroom dropped out or assumed that computing wasn’t a path open to them. “The othering that happens in computer science classes is really a big deal,” she says.
In describing the environment of computer science classes (which then becomes the environment at technology companies), DiSalvo notes that unconscious bias plays a significant role in who is welcomed and who is excluded. “There are a lot of teachers who think some people have the genetic disposition to be good at computer science or not,” she says. “You have the geek gene or you don't.”
While tech companies like Google and Amazon battle over fresh talent, offering six-figure salaries to students coming out of top technology schools like MIT, Carnegie Mellon, and Stanford, historically black colleges and universities have largely gone ignored. Studies from the National Science Foundation and the AIMS Public Health journal confirm that HBCUs are responsible for educating a disproportionate number of degree-holding African American professionals in STEM and awarding close to half of all STEM degrees granted to black women in the US.
The problem is not, as some assume, that too few African Americans are going into the formal and applied sciences to participate in advanced scientific production in large numbers. According to a 2016 Bloomberg Businessweek cover story, “Why Doesn’t Silicon Valley Hire Black Coders?,” more than 20 percent of all black computer science graduates attended a historically black school between 2001 and 2009. But mainstream press articles, the experts interviewed for this article, and message groups for black technologists reveal that companies haven’t been recruiting heavily for artificial-intelligence-adjacent positions from the talent pools at HBCUs.
At the Spelman Innovation Lab, director Jerry Volcy set out to equip the next generation of African Americans for the AI revolution. He is in the process of creating a Center for Innovation and the Arts, the centerpiece of which will be Spelman’s recently launched Innovation Lab, which is designed to blend the arts and sciences in a way that prepares graduates for an evolving economy based on global demand for scientific and technical careers.
In August* 2019, Spelman received a $2 million grant from the Department of Defense to boost its programming in machine learning and STEM. The grant facilitated the creation of the Center of Excellence for Minority Women in STEM. But even with the new infusion of resources, Spelman—and other HBCUs—have a long way to go to catch up with institutions like MIT.
Still, the value of HBCU-based instruction is more than just what’s taught in the classroom—it is also the stories and power of the instructors. Consider the careers of Volcy and his former Innovation Lab codirector De Angela Duff, who currently holds a dual appointment as a professor of Integrated Digital Media at NYU Tandon School of Engineering and as an associate vice provost at NYU.
Duff, 48, was an early adopter of the internet. She turned to programming in the early 1990s as a way to combine her love of photography and design. (She was an early adopter of Mosaic, the pioneering web browser released to the public in 1993.) With an MFA in studio art, a BFA in graphic design, and a BS in textile engineering, Duff has taken an unconventional path—art played as much of a part in her studies as science as she pursued advanced degrees in photography and textiles.
Volcy has taken a similarly unconventional path into the sciences. Growing up in East Orange, New Jersey, Volcy was encouraged by a family friend to study engineering after he replaced the water pump in his mother’s car at age 13. After completing his PhD at the Georgia Institute of Technology in mechanical engineering in 1996, Volcy went to work for Bell Labs in the late 1990s.
But in spite of their accomplishments, both Volcy and Duff found themselves facing an uphill battle with their peers in the broader field. While the environment at HBCUs tends to encourage aspiring technologists from nontraditional backgrounds, both educators have felt the sting of dismissal from people in the business community, and even from advisers on their educational journeys. “There have been many occasions where I've been told overtly that I was stupid,” Duff says.
Volcy breaks in: “Me too. It’s happened to me.”
As inconceivable as that might seem, their experience mirrors those of other African American professionals who have been questioned, isolated, and ostracized at work. With a heavy sense of resignation, Duff shares the advice she offers her own students: “This world is not set up to treat us fairly.”
For Eric Thompson, the Innovation Lab manager at Spelman College, the problem of inaccessibility to STEM fields hits closer to home. A graduate of the Georgia Institute of Technology, Thompson spent his graduate studies trying to build a bridge between the community in Atlanta and the university’s more transient population.
While studying human-computer interaction, he realized that some of the students at Georgia Tech (only around 6 percent of whom are black) barely engaged with the local population. Expressing pride in the city’s strong music and cultural scene, Thompson began a campaign to encourage human-to-human interaction. “I felt like it was almost my duty to tell my peers [to engage with the community],” he explains, noting “they were hardly interacting with what makes Atlanta Atlanta.”
The Atlanta area is rich in opportunity for African American professionals—boasting heavy representation in filmmaking, journalism, technology, shipping, and logistics. But even against the backdrop of new investment, tech funding is another area where African Americans face a tougher road, especially in the early stages of a startup. The racial wealth gap in the United States creates a very different landscape for black entrepreneurs raising the initial round of “friends and family” capital before seeking larger investments.
“As a black individual, our family and friends can’t invest $30K going into a company,” Travis Nunnally says. “As a community, our net worth is at the bottom of the totem pole. People tend to invest in people who look like you.” The Nunnally brothers found some luck and new investors at conferences like TechCrunch Disrupt, but they prefer partners closer to home. And in Atlanta, the investment ecosystem is small and more tightly connected than in Silicon Valley. Travis says he starts with friends, family, and people who they know have industry knowledge, even if they raise less funding overall.
Even if entrepreneurs know they are opting into a harder road, the path into artificial intelligence is tough, even for seasoned professionals.
Abraham Gilbert entered AI in the 2000s after leaving the US Navy. In the service, Gilbert used machine-learning techniques to solve data-based problems in the medical and health sectors. Today he is a data scientist and consultant. He specializes in hospital logistics, using patient data to make treatment more efficient. (Gilbert and the authors of this article are members of the networking organization Black in AI.)
Gilbert likes to think of the possibilities for discovering rare childhood diseases at birth or using deep learning to detect and diagnose diseases like glaucoma without the patient having to wait for a doctor. “People go their whole life not knowing what diseases they have,” says Gilbert, who explains that the capabilities of artificial intelligence could one day make the knowledge of the world’s top doctors and specialists available to everyone, regardless of budget or health care plan.
But he has deep concerns about our current educational system preparing black students for the skills needed to even participate in the global AI-powered economy. “I don’t think this is overt or covert racism. I think we’re just not even on the radar. Our school systems are antiquated and are preparing our students for careers that won’t exist anymore,” he says. “Now we are moving from artificial intelligence toward quantum computing, and you still have kids who can’t use an iPad? We are subjugating minority students to a life of second-class citizenship.”
What is at risk, he says, is nothing less than the future of the planet’s technology. “The democratization of artificial intelligence is this lifetime’s next greatest arms race,” he says. China, he points out, is going into African nations to increase its access to all available resources, including human capital. African nations like Nigeria, South Africa, Ethiopia, and Kenya are developing their own innovation hubs based around AI instruction, but such state-facilitated investments of resources are not being fostered in the US.
While many in the AI field are wary of China’s growing impact in the field, the United States is simply not matching the level of effort China is making in the Chinese provinces or in other nations. Gilbert points out that “children in Kenya [are] now learning [Mandarin] Chinese in the third grade.” In contrast, the United States is allowing its greatest resource—human capital—to go untapped.
The growth of the AI future, both in Atlanta and in the United States more broadly, will largely be a matter of understanding how society functions, how machines work, and how the human world and intelligent systems will have to interact in the future. As Gilbert explains, AI’s untapped potential lies in “having human beings understand other human beings.” In other words, machine learning isn't really about machines. It's about us.
The reporting of this article was made possible through the support of the Eyebeam Center for the Future of Journalism.
*An earlier version of this story misstated the month the DOD grant was received. It was received in August 2019.