Towards the realization of AI research by AI: AI Scientist paper published in Nature journal

2026-03-26 · Source: Blog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Emerging Technologies & Innovation · Depth: Expert, short

Summary

Sakana AI, in collaboration with the University of British Columbia, the Vector Institute, and the University of Oxford, announced the publication of their "AI Scientist" paper in the international academic journal Nature on March 26, 2026. This paper details a system that autonomously completes the entire lifecycle of machine learning research, from idea generation to experimentation and paper writing. The project evolved from an initial preprint in 2024 to "AI Scientist-v2," which produced the world's first fully AI-generated paper accepted through human peer review. The Nature publication provides new insights into the system's architecture, scaling, and the challenges of AI-generated science. Key findings include an automated review system that performs equivalently to human reviewers, achieving 69% balanced accuracy and surpassing NeurIPS 2021 human-to-human agreement rates, and the observation of a "scaling law of science" where paper quality improves with underlying foundation model capabilities.

Key takeaway

For AI Scientists and Research Scientists exploring automated discovery, this Nature publication confirms that AI can autonomously generate peer-review-worthy research. You should consider integrating AI agents into early-stage research to accelerate idea generation and experimentation, while also establishing clear criteria and transparency measures for AI-powered research to address ethical concerns and maintain scientific rigor.

Key insights

AI Scientist autonomously generates research papers, passes peer review, and reveals a scaling law for scientific quality.

Principles

• Research automation is feasible end-to-end.
• AI-generated papers can rival human output.
• Quality scales with foundation model capabilities.

Method

The AI Scientist uses Agentic Tree Search for idea generation, literature review, experimental design, implementation, and execution, then writes papers in LaTeX, with visual models checking figures and tables.

In practice

• Automate research workflows with foundation models.
• Implement automated peer review systems.
• Watermark AI-generated content for transparency.

Topics

• AI Scientist
• Autonomous Agents
• Foundation Models
• Automated Peer Review
• Scientific Discovery Automation

Code references

• SakanaAI/AI-Scientist
• SakanaAI/AI-Scientist-v2

Best for: AI Scientist, AI Researcher, Research Scientist

Related on AIssential

• The AI Scientist: Towards Fully Automated AI Research, Now Published in Nature — AI agents can now autonomously conduct machine learning research, generate papers, and pass human peer review.
• How to build an AI Scientist: first peer-reviewed paper spills the secrets — AI Scientist, an autonomous research tool, successfully generated a peer-reviewed paper, demonstrating AI's capacity for scientific discovery.
• AI-assisted Scientific Discovery, Experimentation, Content Generation, and Evaluation — AI tools are increasingly enhancing the scientific research lifecycle from discovery to peer review.
• How Far Are We From True Auto-Research? — Current auto-research agents generate polished papers but critically lack experimental rigor and result integrity.
• Improving the academic workflow: Introducing two AI agents for better figures and peer review — AI agents can automate complex academic tasks like figure generation and peer review, improving efficiency and quality.
• How AI is changing the nature of mathematical research — AI tools are revolutionizing mathematical research by automating proof generation and contributing novel ideas, accelerating discovery.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Blog.