Curie introduces a framework that enables AI to conduct scientifically rigorous experiments autonomously by embedding methodological best practices into the experimentation process itself.

• Implements three key components to ensure experimental rigor and reliability
• Bridges the gap between LLMs' growing capabilities and the need for scientific methodological control
• Evaluated across computer science domains with promising results
• Provides a foundation for more trustworthy automated scientific discovery

This engineering breakthrough could accelerate scientific progress by automating research workflows while maintaining the methodological rigor essential for valid scientific contributions.

Curie: Toward Rigorous and Automated Scientific Experimentation with AI Agents