Can technology be racist or sexist?
Yes, says Jean J. Ryoo, an education researcher based at the University of California, Los Angeles.
In an effort to address the pressing issues of equity, ethics and underrepresentation in computer science, Ryoo teamed up with fellow researcher Jane Margolis and artist Charis JB to publish the graphic novel "Power On!"
Initially published in the US in 2022, the graphic novel has been translated into Korean by Kim Hyo-won and published here last week by Hangilsa Publishing.
The story centers around four teenage friends -- Taylor, Christine, Antonio and Jon. They are typical teens who communicate via endless texts, share jokes and worry about starting high school. But when a racially biased artificial intelligence system shakes up their neighborhood, they suddenly realize that technology isn’t as neutral as they thought.
Co-author Ryoo said the inspiration for this book came directly from her research with young people.
“The story is not fiction. It is based on over five years in which I interviewed students in public schools about what they see and want in computer science education,” Ryoo said, speaking with reporters in Seoul on Tuesday.
Ryoo, the director of research at the UCLA Computer Science Equity Project since 2017, found that very specific groups of students -- low-income Asian immigrants, African Americans and Latin Americans -- have not been included in computer science education and have been pushed out of job opportunities.
“Women and non-white people make up a very small percentage of the workforce,” she said. “And this means that their voices, their perspectives and their understandings of the world are not being included in technology today, nor in the problem-solving process that these companies generate.”
She added that similar trends are visible here in South Korea. Despite having globally recognized top quality computer science education, women are underrepresented in computer and engineering fields.
The technology built on a lack of diversity leads directly to discrimination and inequality in our daily lives, the researcher said.
“We are seeing racism, sexism and biases against people with disabilities, people with darker skin, immigrants, gay people -- all of these are built into the code that computer scientists create.”
She referred to Canadian computer scientist Joy Buolamwini’s research which found AI face recognition systems have a 1 percent error rate for lighter-skinned men, while a 35 percent error rate for darker-skinned women. The system could not identify Oprah Winfrey, Michelle Obama or Serena Williams, yet these same tools are being used by police officers in the US, according to Ryoo.
Ryoo pointed out another controversial case of a viral video posted on Twitter showing an automatic soap dispenser using infrared sensors that did not dispense soap for a darker-skinned individual, while it worked for a lighter-skinned one. Also, in 2016, Microsoft created an AI chatbot on Twitter, which quickly learned from the internet and became racist and sexist within a matter of hours.
But the issue lies not with technology or AI itself, but with the fact that computer scientists and engineers often perceive their works as “apolitical, neutral, and not responsible for the ethical implication of what they make.”
“Their focus is usually on making the next new thing … ethics are rarely a focus in computer science education or tech companies,” said Ryoo.
The researcher added she hopes this book can be a “conversation starter” that initiates open discussions about how technology intersects with real-world issues.
“Computer science and computer science education can no longer hide behind the wall of false neutrality,” she said.
