Julian Posada spent the COVID lockdown researching the working conditions of Latin Americans who do much of the data work needed for the growing artificial intelligence industry. And he became increasingly frustrated that discussions about the ethics of AI always focused on its uses, and never the poor conditions under which it is produced.
The larger system that supports data production in places like Venezuela, India, and the Philippines — what Posada calls “data extractivism” — and how it affects the quality of that data is now the subject of a book he is writing.
A native of Colombia, Posada is an assistant professor of American studies in Yale’s Faculty of Arts and Sciences, and has affiliations with the Center for Race, Indigeneity, and Transnational Migration, Yale Law School’s Information Society Project, and the Yale Institute for Foundations of Data Science.
We caught up with Posada for the latest edition of Office Hours, a Q&A series that introduces Yale newcomers to the broader university community.
Title | Assistant professor of American Studies |
---|---|
Research interest | How technology is used within different historical, cultural, and social contexts |
Prior institution | University of Toronto |
Started at Yale | July 2023 (Ladder Faculty) |
You have degrees in the humanities and sociology, but your Ph.D. is in information science. What caused you to cross over into that area?
Julian Posada: I wanted to not only study people, but also study how technology is made. What’s a field that allows me to both look at technical stuff and keep the people angle? Information science was the answer.
Tell me about your current research focus.
Posada: I came across this work by first doing quantitative research on this industry. I found that most of these workers were in Latin America.
Venezuela is considered a “cheap” source of labor for the data production industry not just because it has low wages. It’s also because there’s an infrastructure that the government built that can be tapped, and there’s a society that’s very family-based, very community-based, so there’s this social structure you tap into as well. You have people working 12-hour days because there’s the wife taking care of the kids, and the aunt cleaning your house. Also, you have crises where there’s no electricity or water and so on, but still you have a community that brings water into the neighborhood, disposes of garbage, et cetera.
So it’s a system — a territory — that together allows that data production process.
What are some of your findings?
Posada: One is that there is an assumption that if we want to have unbiased data, then we cannot have workers tampering with it or telling us how it is constructed. But by reducing the agency of these people and telling them to shut up, it means that only everything that the client says goes into the data. Workers have views that should also be taken into consideration. If you bring them in, they bring positive things to how the data is created.
You’re teaching a course with another faculty member this semester on the power dynamics related to data, correct?
Posada: Yes, “Introduction to Critical Data Studies,” with Madiha Tahir. We’re loving it because we have students from everywhere: English, Global Affairs, Mathematics, Computer Science, Astronomy. My students were born in the early 2000s — everyone relates to this. They ask very interesting questions about privacy, about who assembled my computer, who is my data being sold to.
What do you enjoy outside of work?
Posada: I love playing video games! I also love eating pizza — I’m a Sally’s guy. And I love swimming.