Computer Science Research Gender Gap Won’t Close for 100 Years

SAN FRANCISCO — Women will not reach parity with men in writing published computer science research in this century if current trends hold, according to a study released on Friday.

The enduring gender gap is most likely a reflection of the low number of women now in computer science, said researchers at the Allen Institute for Artificial Intelligence, a research lab in Seattle that produced the study. It could also reflect, in part, a male bias in the community of editors who manage scientific journals and conferences.

Big technology companies are facing increasing pressure to address workplace issues like sexual harassment and a lack of representation by women as well as minorities among technical employees.

The increasing reliance on computer algorithms in areas as varied as hiring and artificial intelligence has also led to concerns that the tech industry’s dominantly white and male work forces are building biases into the technology underlying those systems.

The Allen Institute study analyzed more than 2.87 million computer science papers published between 1970 and 2018, using first names as a proxy for the gender of each author. The method is not perfect — and it does not consider transgender authors — but it gives a statistical indication of where the field is headed.

In 2018, the number of male authors in the collection of computer science papers was about 475,000 compared with 175,000 women.

The researchers tracked the change in the percentage of female authors each year and used that information to statistically predict future changes. There is a wide range of possibilities. The most realistic possibility is gender parity in 2137. But there is a chance parity will never be reached, the researchers said.

Other science fields fared better. In biomedicine, for example, gender parity is forecast to arrive around 2048, according to the study. About 27 percent of researchers in computer science are women, versus 38 percent in biomedicine, according to the study.

While the study focused on research published in academic journals, the trends may apply to the technology industry as well as academia. Companies like Google, Facebook and Microsoft that are working on A.I. are publishing much of their most important research in the same journals as academics.

Academia is also where the next generation of tech workers is taught.

“This definitely affects the field as a whole,” said Lucy Lu Wang, a researcher with the Allen Institute. “When there is a lack of leadership in computer science departments, it affects the number of women students who are trained and the number that enter the computer science industry.”

The study also indicated that men are growing less likely to collaborate with female researchers — a particularly worrying trend in a field where women have long felt unwelcome and because studies have shown that diverse teams can produce better research.

Compiled by Ms. Lu and several other researchers at the Allen Institute, the study is in line with similar research published by academics in Australia and Canada. While gender parity is relatively near in many of the life sciences, these studies showed, it remains at least a century away in physics and mathematics.

“We were hoping for a positive result, because we all had the sense that the number of women authors was growing,” said Oren Etzioni, the former University of Washington professor who oversees the Allen Institute. “But the results were, frankly, shocking.”

Other research has shown that women are less likely to enter computer science — and stick with it — if they don’t have female role models, mentors and collaborators.

“There is a problem with retention,” said Jamie Lundine, a researcher at the Institute of Feminist and Gender Studies at the University of Ottawa. “Even when women are choosing computer science, they can end up in school and work environments that are inhospitable.”

Many artificial intelligence technologies, like face-recognition services and conversational systems, are designed to learn from large amounts of data, such as thousands of photos of faces. The biases of researchers can easily be introduced into the technology, reinforcing the importance of diversity among the people working on it.

“This is a problem not just when it comes to choosing the data, but when it comes to choosing the projects we want to tackle,” Ms. Wang said.

The Allen Institute study adds to a mounting collection of research pointing to the challenges women face in tech. A recent study of researchers exploring “natural language understanding” — the A.I. field that involves conversational systems and related technologies — shows that women are less likely to reach leadership positions in the field.

“There is still a glass ceiling,” said Natalie Schluter, a professor at IT University in Denmark who specializes in natural language understanding and the author of the study.

Source Link

Get more stuff like this

Subscribe to our mailing list and get interesting stuff and updates to your email inbox.