This piece is a part of The Objective’s Science Series and is funded by the National Association of Science Writers’ Peggy Girshman Idea Grants.
Here’s something you won’t read in the New York Times: Rumman Chowdury is one of the pioneers of responsible artificial intelligence.
Instead, you’ll see Chowdhury described as a researcher who built a tool. Or you might skim over her name entirely. She only received a passing mention.
In July 2021, Chowdhury and her colleague Liz Sullivan spent more than an hour on the phone with Times reporter Cade Metz for a story on removing bias from A.I. systems.
When the article came out, however, only Sullivan was spotlighted, while Chowdhury’s contributions were downgraded to a single line. Her role as the founder and former CEO of Parity, the AI company featured in the article, was not acknowledged.
When I spoke with Chowdhury over Zoom, she recounted analogous experiences with other major media outlets. Chowdhury can name several prominent scientists and technologists, all of them women of color, who have received similar treatment by journalists. Each time, the pattern is disappointingly familiar: women of color are called upon for their time and expertise, then find their names and/or contributions erased from the final product.
“The ask of us is more, and the credit to us is less,” said Chowdhury. “I don’t know of a single male professor that’s spent three hours on the phone and did not get an article written about them.”
When asked for comment, a spokesperson from the Times said, “Our story, which is based on interviews with business owners, researchers, policy experts and trade organizations, provided a look at the growing industry that aims to address problems of bias in artificial intelligence. We’re confident it is a fair and accurate examination of the industry.”
In journalism, it’s common to speak with a number of sources as a story develops. Some of these conversations may be “on background” in order to help the journalist get the lay of the land and decide which direction to take with their story. When consulting people “on background”, the general perception among journalists seems to be that these sources of background information do not need to be acknowledged, especially if there are no direct quotes included in the final piece.
And often, women of color find their quotes on the chopping block, even when their ideas remain.
Where do ideas come from?
Joy Buolamwini, a Black computer scientist, was the first person to call public attention to the issue of biased facial recognition back in 2017. She gave a viral TED talk, describing how facial recognition algorithms struggle to accurately recognize Black faces.
It’s that research that landed her on the CBS News program 60 Minutes.
Except without her name attached. In May 2021, 60 Minutes ran a 13-minute-long feature about racial bias in police departments’ use of facial recognition. The day after the episode aired, Buolamwini tweeted about her involvement with the feature.
Tawanna Petty, a Black grassroots organizer who led the fight to ban facial recognition in Detroit, added that she had also consulted with the 60 Minutes news team before ultimately being cut from the story.
Instead, 60 Minutes chose to feature Patrick Grother, a white, male computer scientist, who works at the National Institute of Standards and Technology (NIST), a U.S. government body tasked with evaluating and ranking the performance of facial recognition algorithms.
In late 2019, Grother’s team published a study that quantified how facial recognition algorithms are less accurate when attempting to identify Black, Asian, and female faces. In the report, they cite Buolamwini, whose work came out more than 2 years earlier, as one of the motivations for their research.
There is no doubt that biased facial recognition is a problem. 60 Minutes was right to call attention to this issue and highlight the stories of people who have been directly impacted by the flawed technology. But when Anderson Cooper lauded Grother and his team for their “landmark study” without mentioning the work of the Black female computer scientists who came before him, Joy Buolamwini and her colleagues were effectively erased from the research agenda that they created.
As an undergraduate working on social robotics, Buolamwini discovered that facial recognition algorithms couldn’t detect her face. Several years later, when she came to the Massachusetts Institute of Technology for graduate school, she encountered the same problem: Facial recognition could only detect her face when she put on a white mask. These personal experiences inspired her Master’s thesis project, which investigated facial recognition algorithms built by Microsoft, IBM, and Amazon and found that they were less effective at identifying the faces of women and people with darker skin tones.
Scientific facts cannot be separated from the context of their production or the identities of the researchers who discovered them. Without Buolamwini’s personal experiences of being undetected by facial recognition, researchers may have never realized that these algorithms can be riddled with bias.
Black women in particular are often among the first to push science in new directions since they contribute perspectives that are outside the norm. As feminist scholars of science like Sandra Harding and Donna Haraway have argued, we bring our whole selves to the table when we conduct research.
The problem of unattributed ideas isn’t unique to science journalism. In 2019, The Lily reported that “female academics aren’t credited in media ‘all the time’,” highlighting the stories of five female historians who said their academic arguments and painstaking research were co-opted by journalists.
Literary scholar Silke-Maria Weineck made a similar point in an op-ed for the Chronicle of Higher Education. Weineck recounted how she and her male co-author were interviewed together for NPR. In the end, only his voice was included in the broadcast; worse still was that her name was not mentioned at all, even in direct references to the book she co-wrote.
I know all of this not just because other women of color have told me. I know this because I have experienced this myself, as have a number of graduate students and early-career scientists I spoke with who asked to remain anonymous out of fear of professional consequences.
In one specific instance, a male journalist adopted some of my exact phrasing (including the title of my research presentation, which he attended) without credit.
I am sure this journalist believed he had done no wrong. He added me on Twitter that week. Imagine my surprise when I scrolled down his feed to see my own intellectual argument, the product of three years of research, nonchalantly presented as if it was a basic truism.
The norm in academia is to cite fellow scholars when building off their ideas. If you borrow someone’s argument, draw from their analysis, or repurpose their problem-framing, it’s expected that you include a reference to their work.
Citational practices differ somewhat in journalism, where a source is typically attributed only when the journalist has included a direct quote or insider information. I believe this disconnect in citational norms is why academics feel so disgruntled when their names aren’t included in stories, while journalists don’t necessarily consider the exclusion an ethical violation.
I was told confidentially by a PhD student that after several negative experiences, she now ignores any messages from journalists in her inbox.
“It feels extractive,” she said. “It’s simply not worth my time.”
What are we losing when women of color opt out of participating in science journalism?
Why does this happen?
In 2015, the nonprofit group Sense About Science commissioned a survey of 218 science journalists. In their report, they detailed a number of reasons why people get cut from stories: tight deadlines, editorial decisions, quotability, or a shift in the direction of the story.
All of these reasons are legitimate. They are also bound up in existing frameworks of power.
For instance, when choosing which sources to quote, journalists may subconsciously reproduce social biases of what a scientist looks (and sounds) like. If you’re prioritizing leadership or seniority when selecting an authoritative source, you’re more likely to end up talking with a white man, because our scientific institutions reflect historical inequalities, especially at the top strata.
Or, as University of Rhode Island journalism professor Kendall Moore put it: “Journalists perpetuate this structure…Whiteness signifies legitimacy and authority, Blackness and Brownness do not.”
Science journalists Adrienne LaFrance and Ed Yong have spoken frankly about their struggle to diversify the expert sources in their reporting. After analyzing a year’s worth of articles, both LaFrance and Yong were dismayed to discover a major gender imbalance in their coverage, despite their personal values.
As a female journalist, LaFrance was distressed to find out that she was part of the problem: “I’m one of the forces actively contributing to a world in which women’s skills and accomplishments are undermined or ignored, and women are excluded.”
Both LaFrance and Yong have said they are committed to reversing these gender disparities. Databases like DiverseSources and Gage have also emerged to assist journalists in locating science experts from underrepresented groups.
“There’s a growing awareness of what it’s costing journalists to have a bias in sourcing, a bias in coverage,” said Erika Check Hayden, director of the Science Communication Program at the University of California, Santa Cruz. “It’s leading to huge blind spots.”
Hayden emphasized the need for concrete incentives if we want to create a permanent change in journalism culture, which is why she now requires students to interview at least one source with a “unique perspective” as part of their assignment grade.
In addition to including more diverse sources, journalists must also consider the contexts in which those voices are highlighted.
Moore, the URI journalism professor, pointed out that people of color often appear in science reporting as victims, not as experts. In the 60 Minutes story about biased facial recognition, the only Black voices included were from men who had been misidentified and wrongly arrested due to faulty algorithms. While it’s important to recognize that technological biases affect real people in the world, media representation should also uplift people of color as experts and changemakers, not only recipients of harm.
In male-dominated industries like science, tech, and academia, it is especially important for women of color to be highlighted in the press. If their names are not featured, this can have a direct impact on their future careers, as they lose ownership over their own work and their legacies are erased. It can also perpetuate the notion that there are not enough qualified women of color working in these industries (an argument that is often used to justify the lack of diversity in upper management positions).
Chowdhury, the tech founder and algorithmic auditor, suggests that part of the reason people get erased from stories is science journalism’s overreliance on hero narratives.
That framing follows a lone individual on their path to scientific discovery, ignoring how scientific and technological progress is often a collective endeavor. Only one person can be highlighted in a hero narrative, which means others’ contributions get pushed out of the story.
Journalism has its own hero narrative, of the lone reporter who uncovers the story and speaks truth to power. It’s an imagination of dogged individualism, which has little to do with the way stories are actually developed; journalism would not be possible without intellectual collaboration.
With that in mind, I want to emphasize that the ideas in this piece were developed in conversation with Rumman Chowdhury, Erika Check Hayden, Kendall Moore, and others who wished to remain anonymous. These women pushed my thinking in new directions, for which I am grateful.
As writers and journalists, my hope is that we can move away from individual bylines and imagine other ways of thinking, being, and discovering as a collective.
Chowdhury, too, encourages journalists to abandon the breathless hero narrative. Instead of raising up certain figures as rebels and revolutionaries, she said, we should turn our attention to collective movements.
“It’s self-defeating if we as underrepresented groups do not push for the collective narrative, because we are never privileged in the hero narrative,” she said.
Nina Dewi Toft Djanegara is a writer and PhD candidate in the Department of Anthropology at Stanford University.
This piece was edited by Janelle Salanga.
The Objective only operates with support from readers like you. Support us today and get your donation matched up to 12x!