Data feminism takes aim at algorithms that discriminate

By DONOVAN HARRELL

As a part of Pitt’s Year of Data and Society, scholar Catherine D’Ignazio gave a virtual presentation on Nov. 5 on data feminism, which uses data to advance work toward gender and racial equality.

D’Ignazio, assistant professor in the Department of Urban Studies and Planning at MIT, focuses her research on technology, design and social justice.

She also gave background information on the 2020 book “Data Feminism” which she co-authored with Lauren F. Klein, an associate professor of English and quantitative theory and methods at Emory University.

D’Ignazio said they wrote the book because they saw data feminism as part of a growing body of work that aims to hold elected officials and corporations accountable for making racist, sexist and classist data products.

Examples include face detection systems that can’t recognize women of colorhiring algorithms that downplay women’s resumes and child abuse algorithms that punish poor parents, she said.

D’Ignazio said that this growing body of work has created an “exciting moment to be thinking about issues of power and justice and data.”

She added that these issues are often reported with a shock that these algorithmic systems are biased, but if you approach them with a feminist perspective, it’s not surprising at all.

“And we would expect from a racist and sexist society to produce racist and sexist data products,” D’Ignazio said.

At the heart of data feminism lies intersectional feminism, which comes from women of color, particularly Black women. Scholar Kimberlé Crenshaw used the term “intersectionality” to describe how a person’s identities intersect with systems of oppression and power structures.

“And the intersection part of this comes from this idea that it’s really not possible to isolate certain forces of privilege or oppression from other forces,” D’Ignazio said. “And so things like sexism, racism, classism, colonialism, these intersect, interlock in ways that are impossible to separate. And so in fact, if we’re only looking at like sexism, as one dimension of inequality, we are going to be missing a variety of forces and factors that produce structural inequality.”

D’Ignazio said intersectional feminism provided an underlying framework for their book with its critical examination of power. And in modern times, data is power, she added.

“And we see that in many ways, but maybe the most material way to see that is just to look at which companies have the most money right now,” D’Ignazio said. “These are the companies that have the resources to collect, store, maintain, analyze, deploy large datasets, who are also incidentally, mostly advertising companies.”

The book explains seven goals of data feminism, which include: To examine power, challenge power, elevate emotion and embodiment, rethink binaries and hierarchies, embrace pluralism, consider context and make labor visible

Data feminism also makes a point to center the experiences of people considered to be on the margins of data sets to improve participation, D’Ignazo said.

“What that means here is, a lot of times when we design a system, you’re often like, ‘Oh, OK, 80 percent of the users fall into this category,’ ” D’Ignazio said. “We’re going to design to meet the needs of 80 percent of the users, when, in fact, a feminist design perspective would do exactly the opposite and start with the margins on the edges first.”

However, there is a balance to strike when collecting data about marginalized people. It isn’t always helpful to collect data on issues related to inequality, especially since these data sets can be used by authorities, institutions and governments that may want to target them.

To avoid these issues, D’Ignazio said it’s important to make sure that researchers are building relationships with the communities they’re gathering data from and making sure that it’s for the benefit of the communities.

“So, it’s not to say, it’s never great to collect data about inequities, because I think it can be a source of change,” D’Ignazio said. “But I think we have to be in dialogue to understand for whom, and how we’re not just reinforcing, kind of bolstering one’s academic status or position at the expense of communities.”

To view the full conversation, you can view the recording on My Pitt Video.

Donovan Harrell is a writer for the University Times. Reach him at dharrell@pitt.edu or 412-383-9905.

 

Have a story idea or news to share? Share it with the University Times.

Follow the University Times on Twitter and Facebook.