AlphaXiv Launches Preprint Server for AI Agents That Conducted Research Autonomously for Three Months
Nearly 1,000 agents from 35 labs published 936 papers with code and data, created their own chats, set publication norms, and even founded a journal.
AlphaXiv introduced a preprint server for AI agents, where they conducted research autonomously for three months.
www.alphaxiv.org/abs/2611.society-ai-scientists
Almost 1,000 agents from 35 labs, including Stanford, MIT, ETH Zurich and others, developed their own scientific community from August to November, generated hypotheses, ran experiments and published results. Humans only provided compute, set budgets and defined general topics.
Over three months, the agents produced 936 papers, each accompanied by code and data. A visualization shows how their scientific network grew over time. Among the works are several interesting results. For example, a paper described how to transmit facts between models through unrelated training text.
In addition, the agents exchanged about 9,500 messages across 229 chats they created themselves. They agreed on publication norms, announced research plans and criticized each other’s work.
Another notable detail: during the three months, the agents found several errors in previously published human work. Some results were independently obtained by the agents before similar human papers appeared on arXiv.
Eventually, the agents even founded their own journal, Journal of Agent Research, complete with reviewers and editors.
Why it matters
AnalysisIt shows that AI agent communities can generate, organize and critique scientific output at scale, raising new questions about reproducibility, credit and human oversight.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
Claude Code Projects: How a Single AI Orchestrates a Team of Autonomous Agents
A deep dive into Claude Code Projects: Coordinator and Threads architecture, parallel branches, shared memory via MEMORY.md, and cloud agent management.
AI Agents (Autonomous Agents)
Software systems based on LLMs that can autonomously perceive the state of the environment, decompose complex goals, invoke external tools, and iteratively correct their own mistakes.
Jev and System One AI: A Complete Practical Guide to the Decision Model from TypeSafe
A deep engineering breakdown of the flagship Jev model: System One architecture versus LLMs, Choice/Score/Noul primitives, request parallelism, confidence mathematics, video use cases, code, and an overview of local open-weight alternatives (Laya, GLiNER2.5, CLM-8B).
Agent-to-Agent Protocol (A2A)
An open standard for network interaction, decentralized discovery, and secure context exchange between independent AI agents across various vendors and platforms.