July 8, 2022

Meta Research PhD Fellowship Spotlight: Solving dynamic problems to reduce algorithm bias

By: Meta Research

Each year, PhD students from around the world apply for the Meta Research PhD Fellowship, a program designed to encourage and support promising doctoral students who are engaged in innovative and relevant research related to computer science and engineering. Fellowship recipients receive tuition funding for up to two years to conduct their research at their respective universities, independently of Meta.

As a continuation of our Fellowship spotlight series, we’re highlighting a 2021 Meta PhD Fellow, Andrés Cristi. Andrés is a PhD candidate at University of Chile, advised by José Correa and Paul Dütting. His research focuses on the interplay between optimization and incentives, or situations where the outcome depends on the actions of strategic agents. Fascinated with understanding and designing algorithms, Andrés dates his interest back to the early days of his undergraduate studies in mathematical engineering, which eventually led to solving dynamic problems with the potential to improve people’s lives — something he’s deeply passionate about.

“I want to make things better, and that motivates me,” he says. “While I was completing my master’s thesis, I had the opportunity to work with a group at the university that was in charge of implementing a nationwide school admission system in Chile. The project helped assign children to nearby public schools, and we needed to ensure it was fair and efficient. After having this experience, I was certain this was a space I wanted to work in.”

He was so certain about this direction that he stopped at nothing to make his dream a reality. After pursuing the Meta Research PhD Fellowship in 2020 and not making it to the final round, he applied again in 2021 and was accepted. This perseverance continues to serve Andrés as he wrestles with more complex problems in his research.

Today, Andrés is studying situations where there are limited resources to distribute among a set of agents, helping make decisions in real time. “We often design a mechanism with a specific objective in mind, such as maximizing welfare or revenue, or allocating things in a fair manner,” he says. “But we also need to consider the agent’s incentives to understand how they’ll interact with said mechanism and understand how this may impact our objective.”

While many people may believe that computers make perfectly rational and unbiased decisions, Andrés says there’s still a lot of innovative work to do in his field. Reducing bias in real-time allocation problems is an area he’s excited about. “Algorithms are just a way of automatizing decisions, which means a designer’s bias can be passed on to the algorithm,” he explains. “My work is centered on understanding the fundamental aspects that drive decision-making when we are allocating resources in real time.”

“I’m still figuring out exactly what I want to do in the next few years, but I know this research will be a focus for me,” Andrés says. “When I think about my longer-term plans, I’d love to hold a position at a university.”

Whether he’s doing research at a university or in industry, Andrés says a strong desire to solve meaningful problems will be at the heart of whatever he does. “When I see a problem, I can’t stop thinking about it — even in my sleep!” he says, laughing. “But research is like this. Get obsessed with a question and try to solve it. You might find multiple answers, some of which are interesting and some that aren’t. This is all part of the process. Love it, enjoy it, and don’t give up.”

To learn about Andrés and his research, visit his profile. For more information on award details and eligibility, visit the program page.