In a sparse lecture room at Stanford University, six students are rehearsing a presentation they'll later give to a roomful of VIPs from the university's artificial intelligence lab. Papers are strewn across the table. Hoodies hang over the cloth-covered cushion chairs. One student wears a pair of Pi earrings. Another wears a t-shirt that reads: "i: Be rational! π: Get real!" A sheet of white poster board sits in the corner, with a few words scrawled in black marker. “Monitoring Hand Sanitation in Hospitals Using Computer Vision," it says.
After a while, they give it a dry run. Scripts in hand, the students describe the images they captured from cameras mounted above hand dispensers at Stanford’s Lucile Packard Children's Hospital, and they explain the machine learning techniques they've used—including something called "climbing the hill"—to analyze the footage and automatically determine whether doctors and visitors are practicing proper hand hygiene. Then they queue up the video presentation, and the students really show their age. The – video warns against the dangers of dirty hands using background music from One Direction and Carly Rae Jepsen. As it plays, the girls dissolve into giggles. “I hope people find this funny!” says one fifteen-year-old.
It’s presentation day at SAILORS, the Stanford Artificial Intelligence Laboratory’s Outreach Summer program, the country's first AI summer camp for girls. Backed by more than forty university professors, postdoctoral scholars, and graduate students from the lab, as well as big-name corporate sponsors like Google, the camp aims to remove the Achilles heel of AI research and, indeed, computer science as a whole: there aren't enough women.
The proportion of bachelor’s degrees in computer science earned by women has actually dropped from 37 percent in 1984 to 18 percent in 2014. That's a worrying trend for many reasons, including some that aren't immediately obvious. Research from the Bureau of Labor Statistics shows that by 2020, there will be 1 million more computer science-related jobs than graduating students qualified to fill them. Women and minorities can help fill the gap.
Like other projects—including a similar camp run by Qualcomm—SAILORS believes this must begin early. Before they become women, the camp gives teenage girls the gift of computer literacy—particularly, literacy in artificial intelligence, one of the fastest-growing branches of computer science. In recent years, Internet giants like Google, Facebook, and Microsoft have doubled down on AI, using brain-like systems to automatically recognize faces in photos, instantly translate speech from one language to another, target ads and more, and simpler forms of AI are now pervasive. Amazon uses a form of AI in recommending products you might like on its popular retail site.
At the camp, the gathered ninth-graders not only explore how computer vision can promote hand sanitation. They apply natural language processing to Twitter in an effort to improve disaster relief. They use algorithms to examine DNA patterns and programmed simple robots. The curriculum was designed to dovetail with the Stanford lab’s own AI projects, and the campers rub elbows with notable AI researchers like Professor Edward Feigenbaum—not to mention field trips to the Googleplex and Silicon Valley's Computer History Museum.
This kind of early exposure can make all the difference. According to one study from Google, those who had the opportunity to take an advanced-placement computer science exam were 46 percent more likely to show interest in a computer science major. Sophia Catsambis, a professor of sociology at Queens College, City University of New York, who studies gender and education, says we must ensure that all schools have a strong computer science curriculum, but programs like SAILORS can play a vital role as well—especially in keeping a so-called "summer setback" at bay. "Additional opportunities are needed, especially for disadvantaged students," Catsambis says.
SAILORS was the brainchild of Olga Russakovsky, a recently graduated PhD student who spent eight years at the Stanford AI lab. Russakovsky studied computer vision, focusing on large-scale recognition, under the aegis of her adviser, Fei-Fei Li. In December, Russakovsky approached Li to discuss how she should spend her last few months at Stanford, and their chat spawned the idea of a summer camp for girls.
Russakovsky had attended the Stanford math camp as a teenager, later serving as a counselor, and she knew how much these summer camps could influence young students. Her idea was to encourage participation from girls and minorities in particular. "If we’re trying to give more exposure to girls and underrepresented minorities, if we’re trying to improve the landscape of AI, we should do it here," she told Li. And Li enthusiastically agreed, saying she'd been wanting to do something on the issue for years.
Working alongside Rick Sommer, the director of Stanford’s pre-collegiate summer institutes and one of her mentors at Stanford Math camp, Russakovsky pulled together a group of volunteers—including her own mother—and started advertising on the university website. Within three days, 300 people had started applications. So the school quit advertising, and eventually accepted twenty-four into the program. Applications were open to any rising high school sophomore, and targeted girls, but the group exhibits some added diversity. Most of the participants are Asian or South Asian, and two are white.
A daughter of two math teachers, Russakovsky says she doesn't subscribe to the belief that it is boys who naturally gravitate towards science and technology subjects, but at the same time, she acknowledges that it's become a complicated issue. "Just because you don’t want to be in a group of people who are all different from you where you stick out, that doesn't necessarily mean you’re not interested in math and science," Russakovsky says. "We need to somehow decouple the two."
As a teenager and on into college, Russakovsky says she had many doubts about being a girl and a woman in the world of mathematics, but that, with diligence, she pushed through those doubts. With the camp, she hopes to help other girls with similar problems, and help them get "unstuck." "I’ve seen people leave the field, or pivot slightly to something else,” Russakovsky says. “They will leave companies. They will leave programs. At some point, you just get tired. It’s not that you’re tired of computer science or math, it’s just that you realize you can choose to be in a different environment."
Li, her mentor, says much the same. But she also emphasizes how much the industry needs these young girls. "Today, engineers writing the algorithms need to be aware of ethical, social issues," Li tells WIRED. "It’s really important to bring that education and perspective into our research. We need to deliver that mission statement, and we need to attract people who care."
One August day, the twenty-four campers trek to Mountain View and the Googleplex, the headquarters of the Internet's most powerful company. They're here for a panel talk with three female professionals working in AI, and their excitement is palpable. “Clap once if you can hear me!” Russakovksy yells out when the chitchat gets a bit too loud. "Clap twice if you can hear me." The tactic works, most of the time.
The enthusiasm continues during the panel, as the campers ask how Google AI engineers spend their work days (one product manager's answer: lots of meetings) and explore the pros and cons of working in industry versus academia. They listen with rapt attention as the Googlers describe the work that goes into building a system that can identify photos using natural language; figuring out how to queue up another YouTube video a user will likely enjoy, given a video she has just watched; and the near-magic of on-the-fly voice recognition and task processing in the Google app. There were quite a few jokes about the wonders of data centers tossed around. And lots of selfies snapped.
Then the girls ask about discrimination. And though the Googlers indicate this isn't much of a problem, the ninth graders later tell me they experience discrimination, in small ways, all the time.
“I’m really aware of being the only cheerleader in my robotics class,” one girl says. “The first day I walked in, all the eyes were on me. They were like: ‘You’re in the wrong classroom.’” Another student, named Anooshree, says she is part of an all-girls robotics team. At tournaments, she says, they often heard two questions: "Is that your entire team?" and "Did your dads build this for you?”
As the stories pour out, one camper expresses her surprise—and relief—after meeting other girls who had experienced many of the same feelings she had. “I never realized this was a shared experience before, not just me in my particular environment,” she says. “I think for a lot of us, it’s nice to have someone to relate to.”
At the same time, they express a real pride in who they are. “I think when everyone was playing with Barbies, we were all making toy helicopters,” says one camper. Another continues: "Yeah, when I would get Barbies as a gift, I would just give away the doll to someone else." But they aren't thinking in black and white. There's some real self-reflection at play. A third camper completes the thought: "I don’t think playing with a certain toy prevents you from being good at computer science."
They all agree on one thing: SAILORS has given them a reason to be confident. And many say they will encourage other students to apply next year—though some question how much success they will have recruiting others who aren't interested in computer science. Still, they say, camp has shown them how fun AI can be and, perhaps more importantly, how it will change technology.
Before we leave the Googleplex, one camper says there may to a trick to bringing other students into the camp. They need to see, she explains, that AI drives things like video games and social networks. They need to see that AI is the future. The best way, another camper concludes, might be to give other kids who haven't been exposed a real "hands-on" experience.
