By Brendan Galvin, CISE Staff
Boston University computer scientist Professor Mark Crovella has spent years studying the unseen mechanisms that shape what people encounter online—from recommendation algorithms on platforms like YouTube to the powerful large language models (LLMs) now driving AI systems. His research explores how these technologies can unintentionally amplify bias and misinformation, influencing public opinion and decision-making at massive scale. What began as an investigation into biased recommendation systems has evolved into a deeper pursuit: uncovering the inner workings of LLMs through interpretability research, which seeks to make their complex computations understandable to humans.
Crovella’s work recently earned him a spot as one of 35 projects selected for the National Artificial Intelligence Research Resource (NAIRR) Pilot, which provides academic researchers with the computing power and data access traditionally reserved for major tech companies. By advancing interpretability in an open academic setting, his goal is to identify where biases exist within AI models—such as OpenAI’s GPT series—measure their impact, and make these processes more transparent. This work could ultimately inform public policy and guide responsible AI governance.
A founding member of BU’s Faculty of Computing & Data Sciences and one of its inaugural Duan Family Faculty Fellows, Crovella continues to shape both research and education in data science. His work bridges technical insight with societal impact, helping ensure that the algorithms behind modern AI systems operate fairly and transparently.
Learn more about his groundbreaking research in this CISE story.