1,200 Researchers Used Coding Agents to Reproduce Over 2,000 ICML Papers
Gradio · x · 2026-08-06
Gradio announced the completion of an unprecedented large-scale reproducibility audit in the AI community.
The project mobilized over 1,200 participants who used coding agents to reproduce papers from ICML 2026. They successfully reproduced or falsified over 2,000 papers, accounting for a third of the entire conference. Gradio will host a livestream tomorrow at 9 AM PT to announce the winners and share their findings.
Related event: 1,200 People Use AI to Reproduce ICML Papers, Over a Third Refuted(3 posts)→
More from coding & agent
- Claude Code v2.1.280+ lets you switch Opus 5.5 effort mid-session without breaking prompt cache — kimmonismus · 2026-09-23
- Sentry founder David Cramer finds Claude Code's new sidebar confusing vs Codex — zeeg · 2026-09-23
- Mid-session effort switching on Opus 5.5 preserves prompt cache in Claude Code v2.1.280+ — lydiahallie · 2026-09-23
- fable-advisor v6: open-source Claude Code plugin teams up Opus 5.5, Fable 5.1, GPT-6 Sol and Luna — daniel_mac8 · 2026-09-23
- Open-Source fable-advisor: Fable 5.1 Architect Orchestrating GPT-6 and Opus — daniel_mac8 · 2026-09-23
- GPT-6 Sol vs Grok 4.7: one prompt, one HTML file, three Star Wars worlds each — rohanpaul_ai · 2026-09-23