Project title: AI Safety and Content Recommendation Algorithms
Project description: With the advent of large language models, AI systems now play a more varied and important role in our lives. From browsing the web and scrolling social media to applying for a job or a loan, we unknowingly interact with these models daily. As such, it is ever more important that these models behave safely and do not develop harmful behaviors. I am primarily interested in exploring AI safety in the context of social media content recommendation algorithms. A common problem among these algorithms is that they lead to the creation of filter bubbles, which isolate users from diverse viewpoints. More alarmingly, some of these algorithms could manipulate user opinions in ways that make their interests and future behavior much easier to predict. I aim to study the prevalence of these harmful behaviors on today’s social media platforms as well as develop quantitative metrics to measure and mitigate these negative effects.
Most important achievement: I am particularly proud that I got to play a key role in designing a COVID-19 hospital in Myanmar during the pandemic.
Reflections on the College Scholar Program: The College Scholars program has offered me the opportunity to study topics that truly interest me. I have also been able to find a community of peers who, despite pursuing different research directions, share a common passion for intellectual exploration.