AI-Scientist-v2 is an intelligent scientific research automation system developed by SakanaAI, Japan, with the core goal of autonomy in the entire process from research ideation to paper writing. The system provides users with an end-to-end research solution by integrating multiple AI technologies.In April 2025, the tool was open-sourced on GitHub and received a major upgrade from the first version, introducing Agentic Tree Search technology.
Key features include:
- Intelligent Research Idea Generation: Automatically generates viable research topics based on user input directions
- Experiment code writing: generate runnable experiment code and support optimization adjustment
- Autonomous experiment execution and analysis: automatically run the code, collect and process data and generate visualization charts
- Academic paper writing: according to the experimental results automatically generated in line with the academic norms of the paper
- Research Path Optimization: Intelligent Exploration of the Best Research Options through Agentic Tree Search Algorithm
- Literature support: access to Semantic Scholar API for novelty checking and citation addition
The tool is particularly suitable for machine learning related fields, and has successfully produced AI-authored papers that have passed the peer review of the ICLR 2025 workshop, signaling its practical application value in academic research.
This answer comes from the articleAI-Scientist-v2: Autonomous completion of scientific research and paper writingThe
English 
简体中文 
日本語 
Deutsch 
Português do Brasil