MIT researchers join AI2050’s 2026 fellowship cohort

MIT’s announcement, published on October 30, names faculty members and alumni among the 2026 AI2050 fellows. The initiative supports research aimed at making AI beneficial to society. Several projects ask how people can use algorithmic insights while retaining a meaningful role in interpreting them.

Public input that decision-makers can inspect

Bailey Flanigan’s project tackles decisions involving complex policy choices or technical systems. The team plans to prototype a way for people to contribute text, comparisons and structured responses. The intended result is an explicit representation of public judgment that can be challenged, giving decision-makers something more informative than an opaque algorithmic answer.

When many people rely on the same models

Manish Raghavan examines how widespread AI reliance could affect markets, democracy, science and journalism. His project description identifies a risk: people using the same models may make less independent judgments and share mistakes. The planned work develops ways to measure that risk and evaluation standards rewarding diverse, complementary insights; it does not report a demonstrated institutional outcome.

From predictions to understandable theories

Sendhil Mullainathan’s team is building a toolkit intended to translate patterns detected by AI into scientific theories people can understand. The motivation is that useful predictions can remain difficult for human experts to interpret. The project presents a research ambition for connecting computational patterns with scientific explanation, rather than evidence that a universal discovery tool is already complete.