AI Scientist by Sakana AI
AI Scientist is an open-source, fully autonomous research agent developed by Tokyo-based Sakana AI. It automates the scientific research process end-to-end—from hypothesis generation and code implementation to experimentation, analysis, manuscript writing, and even peer review. Inspired by the iterative nature of human discovery, AI Scientist has already authored peer-reviewed papers in the machine learning domain.
Key Attributes
End-to-End Automation: Generates novel ideas, writes and executes code, analyzes outcomes, and composes complete academic papers.
LLM-Powered Peer Review: Uses a large language model to simulate peer reviewers and refine outputs through iterative feedback.
Cost-Efficient Research: Produces full-length papers at a fraction of traditional costs (estimated ~$15 per paper).
Open Source: Licensed under Apache 2.0, enabling free use, modification, and commercial deployment.
Collaborative Roots: Developed with contributors from institutions like the University of Oxford and University of British Columbia.
Example Use Cases
Automated ML research on diffusion models, transformers, and related areas.
Iterative ideation for cutting-edge academic exploration with minimal human involvement.
Training and educational tool to teach research methodology and AI capabilities.