Google DeepMind and MIT Unveil CoDaS AI Agent System for Autonomous Scientific Research

Google DeepMind and the Massachusetts Institute of Technology (MIT) jointly developed CoDaS, or AI Co-Data-Scientist, a system in which multiple AI agents divide tasks including hypothesis generation, statistical analysis and result validation. It aims to turn raw data into testable scientific findings while reducing the extensive human labor and expert time required by the research process.

The latest tests showed that CoDaS could identify research-worthy health indicators in wearable-device data and autonomously complete the analysis, validation and paper-writing process. The entire workflow took about 8 hours, sharply shorter than the roughly one month experts previously required. Available information did not disclose a formal release date, investment amount or commercialization timetable.