Gabriel Recchia

Gabriel Recchia is a cognitive scientist and director of Modulo Research, where he works on the evaluation and alignment of large language models and the design of scalable oversight protocols. At Modulo he has led work releasing datasets of expert-annotated valid and invalid long-form solutions for use in scalable oversight experiments and related studies on how humans and models evaluate complex answers. Previously, he led user-testing research and evaluation for patient-friendly genetic reports and the Predict: Breast Cancer prognostic tool at the University of Cambridge’s Winton Centre for Risk and Evidence Communication, and has co-authored widely cited research on risk perception and communication.

Personal Website

https://gabrielrecchia.com/

LinkedIn

-

Twitter / X

-

LessWrong

-

EA Forum

-

Organizations

Projects

No grants recorded.