Mapping the pipeline from robot idea to robot peer review

What it does

This repository accompanies a survey paper on AI auto-research and acts as a curated reading list for the emerging field. It corrals papers and tools into an eight-stage research lifecycle spanning creation, writing, validation, and dissemination. The maintainers also used their own survey as guinea-pig material, generating slide decks and posters with Paper2X pipelines to see what the tools actually spit out.

The interesting bit

Instead of dumping links alphabetically, the project imposes a pipeline taxonomy that forces you to confront which part of science is being outsourced—idea generation, code execution, rebuttal drafting, or turning a PDF into a conference talk. The exhibition section is essentially dogfooding: they fed the survey through GPT-5.5, Claude Opus 4.7, NotebookLM, and Manus 1.6 so you can judge the output yourself.

Key highlights

• Organizes the field into eight stages across four phases, from literature review and coding to peer review and dissemination.
• Tracks verification and assessment subtopics, including code reproducibility, review quality, and AI detection.
• Hosts an exhibition of AI-generated slide decks and research posters produced from the survey paper.
• Covers end-to-end autonomous systems, domain-specific pipelines, and evolutionary self-improving agents.
• Maintains sections on societal impact, critical perspectives, and a dedicated tools directory.

Caveats

• The README is largely a table of contents; substantive analysis lives in the linked paper and project page.
• Several explanatory sections (e.g., “Background” and “Five Central Findings”) are commented out in the source, suggesting the documentation is still being assembled.
• This is a curated bibliography and taxonomy, not a runnable framework or library.

Verdict

Researchers mapping the AI-science landscape should treat this as a strategic index; developers hunting for a drop-in auto-research codebase will not find executable code here.