Seven people with MIT ties have been named to the 2026 cohort of AI2050 Fellows, an initiative that supports work on difficult problems in artificial intelligence. The group includes five MIT faculty members, two alumni and a former postdoc. Their projects connect AI research with questions about people, science and public decision-making.
Research framed around AI’s possible benefits
AI2050 is a philanthropic initiative conceived and co-chaired by Eric Schmidt and James Manyika. Schmidt Sciences, Eric and Wendy Schmidt’s philanthropic initiative, announces the fellows annually to accelerate scientific innovation.
The fellows’ work is shaped by a forward-looking question: “It’s 2050. AI has turned out to be hugely beneficial to society. What happened? What are the most important problems we solved and the opportunities and possibilities we realized to ensure this outcome?” The question asks researchers to consider what useful AI could make possible, and which problems their work might help address.
The MIT faculty projects span several fields. Some focus on how AI systems learn or interpret biological data; others examine how algorithmic tools affect people and public institutions. Taken together, they show the range of research represented in this cohort.
Four faculty projects link AI to real-world questions
Sendhil Mullainathan, an AI2050 Senior Fellow, is the Peter de Florez Professor at MIT, with appointments in Economics and Electrical Engineering and Computer Science. His research centers on the relationship between algorithms and people. His AI2050 project focuses on translating patterns detectable by AI into testable theories.
Alex Damian, an assistant professor in Mathematics with a shared position in Electrical Engineering and Computer Science, studies the mathematical foundations of deep learning. His work emphasizes optimization dynamics and representation learning, while his AI2050 project focuses on effective AI model training.
Bailey Flanigan’s research connects algorithms and social science to questions of democratic participation. She is interested in sampling algorithms, measuring opinions and preferences, and designing democratic innovations such as deliberative minipublics and participatory budgeting. Her project aims to gather structured public input that lawmakers can use when making complex decisions.
Yunha Hwang works across machine learning and biology, with research interests in microbial evolution, biomanufacturing and open science. She is also co-founder and chief scientist at Tatta Bio, a scientific nonprofit dedicated to genomic AI for biological discovery. Her AI2050 project focuses on determining biological enzyme functions.
Fairness and AI reliance are also in focus
Manish Raghavan, a professor at the MIT Sloan School of Management with a shared position in Electrical Engineering and Computer Science, studies computational methods for issues of social concern. His research includes algorithmic fairness and behavioral economics, especially the use of algorithmic tools in hiring.
His AI2050 project examines how to mitigate negative effects and strengthen positive aspects of widespread reliance on AI. That focus places the consequences of using AI systems alongside their technical capabilities.
Mullainathan’s broader work also connects behavioral science and technology. He co-founded ideas42, a nonprofit applying behavioral science, and the Abdul Latif Jameel Poverty Action Lab at MIT, which promotes randomized control trials in development. He also co-founded Dandelion Health, which combines proprietary AI tools with real-world patient data.
Alumni and a former postdoc join the cohort
MIT alumni Irene Chen PhD '22 and Allison Z. Koenecke '14 were also named AI2050 Early Career Fellows. Former postdoc Juan Carlos Perdomo joined them in the cohort.
The announcement brings together researchers working on AI training, biological discovery, democratic input and the effects of algorithmic systems. Each project approaches the AI2050 question from a different direction, asking what research could help make AI beneficial in practice.