Research Interests
Ira Globus-Harris works on the algorithmic foundations of responsible computing. From the news curated on your favorite newspaper app to triaging emergency resources and determining loan eligibility, machine learning shapes our experiences of the world and is here to stay. This comes with potentially powerful benefits: reduced inefficiencies in resource allocation, increased accuracy in important decision-making, and content curated to match your needs. It also has drawbacks: If someone makes an error, you might make a complaint, try the same process again with a different person, or even file a lawsuit. However, these interventions require an understanding of not just what mistake happened but also how they happened. If an algorithm makes a mistake, especially when deployed at scale, it is often unclear how to resolve the problem. Ira’s work broadly looks at mechanisms to resolve these potential harms incurred by AI-driven decision-making. Their work focuses on using algorithmic techniques which are scalable for real-world use, holistically consider an algorithm in its broader context, and which flexibly incorporate human input.