Dream Makers

Daphne Koller, photographed in Insitro’s biomedical laboratories in South San Francisco.
Koller, a MacArthur Fellow, tech entrepreneur (she cofounded the online learning platform Coursera in 2012 and, more recently, the drug development company Insitro), and computer science professor at Stanford University, understands the importance of instilling values early—particularly where expectations are concerned. “We all hear the horror stories about groping and sexual harassment, but even if you put those aside as egregious behavior and you just think about normal day-to-day operations, there are just a lot of areas where women are made to feel less than in the workplace.” Koller says.

“[The inequality] starts from the early stages, with girls not going into math and technology classes,” says Koller, citing studies that show girls excel once the societal barriers are removed. “In PISA tests, which are an international math test that’s run across probably around 50 countries, the x-axis was gender parity in the country as measured by U.N. metrics as aggregate, and the y-axis was the math gap. What was interesting was that the more equal a country was across those other metrics, the lower the math gap was. The point that when you got to Scandinavia, there was no math gap. In fact, girls were slightly outperforming boys.”

A view of Insitro’s laboratories in South San Francisco.

As AI becomes more prevalent and machines continue to learn, ethical concerns have risen surrounding what and who is teaching them. As technical colead of the Ethical Artificial Intelligence Team at Google, Timnit Gebru has uncovered many of the ways that algorithms betray their creator’s biases. “Data is generated by society,” Gebru says. “If you have some biased data and you’re trying to gain insights from it, you can arrive at biased conclusions.” Gebru is wary of the power imbalances that the technology could be used to uphold. “It’s really about humans and what we’re trying to achieve,” she says. “If we have a state where people want surveillance or they’re okay with it, that’s where AI’s going to go. If we have a place where people want to invest a lot of money into education and healthcare projects, that is where it’s going to go.” The scope of the social issues at play can at times feel overwhelming for someone who got into science because first and foremost, they loved science. “I think of it this way," Gebru says. "Physicists did not become physicists so that they can build a nuclear bomb or so that they can have activism against it. But once you are a physicist, then when you start to see some of the things that you think are harmful, it’s your responsibility to lead the movement against it. I got into the field wanting to do certain things, but it’s hard to do the things that are fun when you think about how that [technology] could be misused. I spend a lot of my time trying to figure out how to mitigate that harm.”

Timnit Gebru, photographed at Stanford University.
Timnit Gebru’s role as technical colead of the Ethical Artificial Intelligence Team at Google has made her one of the most visible advocates for diversity in tech and taps into many of the current concerns about AI and accountability. “I'm doing the things that might not be the most fun, but that I think the industry really needs,” she says. She founded Black in AI, a community for fellow researchers of color, with another computer science researcher, Rediet Abebe, in 2015. The group deals with what Gebru calls “intersectional issues, accessibility, class. . . .What does it mean to have all of these international companies mostly in the U.S.? What does it mean for power to be not geographically distributed around the world?”

Though they make up only around 20 percent of the workforce in Silicon Valley at present, the contributions of these women are shaping our collective future. And if you ask them, one of the most exciting developments has been the shift in perception of what exactly constitutes the field they’re in. “A lot of people think that technical fields and especially computing are this kind of geeky thing where people go off and do these obscure mathematical formulas that have no relevance to anything,” says Daphne Koller, Stanford professor, AI researcher, and entrepreneur. “What we’re now seeing is that computational methods are becoming pervasive in so many professions. You can help fishermen in South Africa figure out where to sell their fish or use computing as we’re doing to discover new drugs. There are just so many ways now to have a tremendous impact on society.”

Carol Reiley has spent the past 20 years studying the kind of exchanges that happen when robots are integrated into daily life. "When you figure out where humans are powerful and where humans are vulnerable, how does a robot complement that?" she says. Take her work with the da Vinci Surgical System, which is controlled by a surgeon from a console and designed to facilitate minimally invasive complex surgery. “Its job is not to replace the doctor, but to make the doctor superhuman,” she says. “You have a hand reaching out. You have motion tremor reduction. You can slow down your motions, or you can scale up your motions. That’s what’s cool about robotics: We compliment humans, so that you can, for a moment, have superhuman abilities.” A MacArthur and Guggenheim Fellow and computer science professor at the University of California, Berkeley, Oasis Labs cofounder Dawn Song is most excited about the work her team is doing in “program synthesis learning—essentially training computers to learn how to write code.” She hopes the new technology will open doors for a more diverse range of creators. “Our [current] software systems are very powerful. . .but we still need humans to write the programs,” Song says. “For a lot of users who don’t know how to write code but have great ideas, it’s difficult to take those ideas and implement them in a real-world system. If computers can write code, it makes software development much more efficient and accessible.” It's a whole new world.

Carol Reiley, photographed at Davies Symphony Hall, where she serves on a new creative advisory board for the San Francisco Symphony.
The myth of the solitary creator may loom large in Silicon Valley, but for Drive.ai cofounder Carol Reiley, engineering became a path towards engaging a global community. “It’s almost a way to talk to more people and reach more people than I ever would have,” she says. “The Internet has given us this amazing capability. YouTube is a way for an individual to connect with thousands of people all at once. I think that’s especially why I’m more drawn towards robotics than anything—it’s taught me more of what it means to be human.”

“There is a compassion problem [within the industry],” says Reiley, who was 9 months pregnant during this photoshoot: The AI schools are the elite schools, and “often founders are out to become the next billion-dollar company . . . I’m not religious or any of that, but an ethical foundation should be taught at every school. Especially for those gifted the power of building AI technology. The ethical side and intellectual side have to be balanced, or else we’re going to start seeing bigger problems as we go into this fourth industrial revolution.”

Reiley cofounded Drive.ai in 2015.
While much of the media attention regarding driverless cars has centered on the ramifications for the trucking industry and whether or not the automated SUVs will replace Ubers, Reiley is eager to tackle another issue: drunk driving. “We are, as humans, very distracted, terrible drivers, and we don’t know it.” She says, “94 percent of those accidents are preventable. These are just careless errors, and I think around 1.3 million people die every year from car accidents. When you think about ways to save lives, the impact that self-driving cars can have is overwhelming.”

Devi Parikh, photographed at Facebook’s headquarters in Menlo Park, California.
For many, AI stokes fear about robots replacing human workers, but for Parikh, a research scientist at Facebook AI Research and an assistant professor at Georgia Tech’s School of Interactive Computing, human interaction is still one of the most essential elements of the process. For the robotics field to evolve, people have to serve as the primary teachers. “I think there is a huge gap in terms of what technology can do today versus what people think it can do. We’ve made a lot of progress, there’s no denying that, but we are so far from being able to have any notion of general intelligence,” she says. “If a piece of software can play chess or Go, it can do that one thing very well, but it can do nothing else. . . . It's not just about making things better in isolation, but making it better in a way that complements humans, so that you can have more effective human-AI teams working towards common goals."

Dawn Song, photographed at the University of California, Berkeley.
Dawn Song’s startup, Oasis Labs, developed new blockchain technology (a form of decentralized information distribution that allows messages to be distributed without being copied) that proponents believe could revolutionize the way private data is handled online. “As more important decisions are being made using the existing systems, it’s necessary that we make sure they’re secure,” says Song.

As a professor, Song has seen an uptick in the amount of female students in entry-level science courses, but she’s eager to see more. “We need more women in the field,” she says. Still, her increasingly digitally native Gen Z students are a source of hope. “Increasing the availability of technology has absolutely helped. Now you have high school students who are already very accomplished in AI, who enter university with the kind of experience you couldn’t have access to in the previous generation.”