AI researchers say papers should be executable, verifiable artifacts
AnneliesGamble · x · 2026-07-21
Reproducibility is no longer an afterthought in AI research.
The post argues that when papers can be generated at near-zero cost, gaming human review also becomes cheap. Instead of asking reviewers to catch more errors, the proposal is to submit research as agent-native artifacts:
- record the full research trajectory
- make the work executable
- make the results verifiable
The cited example is a NeurIPS submission where non-technical authors said they used agents for all coding, did not understand the underlying code, and expected reviewers to catch any mistakes.
Related event: AI-Generated Content Challenges Academic Accountability(2 posts)→
More from Research
- FrontierML benchmark shows open models still far behind closed ones on end-to-end ML work — scaling01 · 2026-10-27
- KANLab: an interactive browser playground for Kolmogorov-Arnold Networks — Lukas Niehaus · 2026-10-27
- CMU's LLM2Cert synthesizes stability certificates verified via SMT and Lean — CarnegieMellonU · 2026-10-27
- NVIDIA open-sources FLORA, a robotics data engine ~125x cheaper than GPT-6 Astra — nvidia · 2026-10-27
- RSNA 2026 Deep Learning Labs return: eight hands-on AI classes over four days in Chicago — PRouzrokh · 2026-10-27
- Steering two orthogonal concept neurons makes a model compose "gnome" — Sauers_ · 2026-10-27