The initiative, termed 'Open RSI,' aims to facilitate AI that improves AI in a public environment, using Hugging Face Inference Endpoints to serve models for live feedback. All project artifacts, including the research trajectories, training data, and code used by the agents, will be open-sourced on the Hugging Face Hub once the experiment ends.
Wednesday, October 21, 2026
Story
RSI Arena and Hugging Face Deploy AI Agents to Train Models via Community Voting
AI14 days ago
The initial phase of an autonomous AI research experiment has concluded with the deployment of models that AI agents built without human intervention in data selection or coding. In a partnership with Hugging Face, RSI Arena now employs these agents to develop models that the community judges in head-to-head matchups, feeding real-time voting results back to the agents to guide their subsequent experiments and training.
Looking for more context or analysis?
Tilt chat provides reliable, tailored insights, alerts, and automations.