Graduate Student Fellow Spotlight: Taylor Beauvais, Sociology, CAS
Taylor Beauvais is a third-year PhD candidate in Computational Humanities, Arts & Social Sciences studying Sociology and a 2024 Hariri Institute Graduate Student Fellow. His research area is on the impact of artificial intelligence in economics, social, and governmental systems. Beauvais is advised by Joan Donovan (COM), Jonathan Mijs (Sociology), and Neha Gondal (Sociology). He has a masters degree in public opinion, research methods, and data analytics from the University of Lausanne and a bachelor’s degree in psychology and sociology from the University of Buffalo.
The Hariri Institute asked Beauvais about his research focus, recent developments in the area, and the impact the Graduate Student Fellowship will have on his work:
Hariri Institute: Can you describe your research focus and its applications?
Beauvais: My research focuses on the root epistemological causes of AI bias and algorithmic harm from a sociological perspective rather than a methodological one. There are innumerable excellent papers and researchers investigating debiasing techniques for these tools, but I think we can only stretch our methods so far. When we try to build AI free from prejudice like sexism and racism, we’re teaching it how we wish the world was, instead of how it is. In application, I’d like to ask what if we allowed our systems to know how prejudicial the world is, and instead of weighing against it, we used it as a chance to intervene against systemic forces?
Hariri Institute: How did you become interested in this? Was there something that inspired this area of interest?
Beauvais: In 2015, freshly graduated from my undergraduate degree in Psychology and Sociology, I was questioned by the police while sitting in my car jam packed full of stuff from a recent move. I was parked in a nice neighborhood in Buffalo, New York and it was about 2 in the morning. It looked odd for a 21-year-old to be sitting in an old well-worn car full of stuff in a neighborhood where people had private tennis courts. Nothing came of it, but my manager at my job told me the next day about an algorithmic system they have for judging recidivism risk. On a whim I looked it up and ran my own profile through it. The results were stark: I was apparently at an extremely high risk of committing future crime. It became a hobby over the next 5 years to keep track of how wrong most of these authoritative systems were.
Hariri Institute: What are the main goals or objectives of your research?
Beauvais: I have a lot of nostalgia about technology and its utility. It seems like the zeitgeist has turned from excitement to pessimism in regards to technology over the past 20 years. I want this one to be done right. I want technology that makes life easier for everyone at the expense of nobody. I want this technological system to counter systemic prejudices instead of complimenting them. I want to get computer and data scientists to think about their projects sociologically, and sociologists to think about their work technologically.
Hariri Institute: Has there been a recent development or finding that you find particularly exciting?
Beauvais: The nature of my work (auditing technological systems for prejudice) isn’t always a happy topic but the findings can be useful in exciting ways. I’m presenting a paper at the American Sociological Association that demonstrates how the primary demographic that informed ChatGPTs knowledge of politics, was young white independent-leaning men. The paper calls into question the actual utility of the transformer algorithm for generalizable models.
Hariri Institute: What advice do you have for students entering the first year of a PhD program?
Beauvais: Protect your down time! You’ll surely hear tropes about long work weeks and intense deadlines and demanding projects, and it may compel you to forgo some of the little pleasures you have in life but those are the things that get you through. Have the coffee with a friend before diving into the analysis (it can wait 10 minutes). Do your writing from a hammock on your porch, not your stuffy desk (you might as well be comfy when writers block hits). Have dinner with your cohort, you all have time to eat (no one needs to skip meals for lab work).
Hariri Institute: How do you plan on using this fellowship opportunity?
Beauvais: This opportunity is going to be a launching point for a technical project I’m still dreaming up. I need some major technical assistance and a good bit of funding to accomplish it so the resources from this fellowship will be vital.
The Hariri Institute for Computing’s Graduate Student Fellows program recognizes outstanding PhD students who are pursuing computing and data-driven research at Boston University.
Learn more about current Graduate Student Fellows and the program here.