研究库 论文知识库

内容库 / 论文

arXiv:2609.09219 · Agent 智能体

Scores Alone Do Not Prove Discovery: The Discovery Certification Protocol for Auditing AI Research Agents

Scores Alone Do Not Prove Discovery: The Discovery Certification Protocol for Auditing AI Research Agents

类型：arxiv
标识：2609.09219
链接：https://arxiv.org/abs/2609.09219
主分类：agent
形态：method
TLDR：AI research agents combine prior knowledge, public sources, and experimental feedback to produce useful results. The Discovery Certification Protocol (DCP) turns claims about these results into executable recovery and feedback tests. Gate 1 validates useful improvement on sealed evaluation. Gate 2 gives matched agents the registered starting information and observed Web content while withholding the target research history. Every valid method reaching the numerical target supplies a recovery witness and triggers the Core veto. DCP Core requires adequate controls, zero observed recoveries, and
待LLM分类：否
来源文件：
/inbox/tom/_candidates/2026-09-10-agent-rag-longcontext-candidates.json

相关 · 按语义匹配

笔记 LLM Agent Credentials Leaked by Skills — ASE 2026

仓库 VoltAgent/awesome-ai-agent-papers

笔记次精读：Argus — Uncertainty Quantification for Computer-Use Agents（短评）

文档 Stephen-on-spark-2026-07-16

仓库 K-ALOHA/cdi-policy-verification

仓库 JeffffffFu/Awesome-Differential-Privacy-and-Meachine-Learning

arXiv

2609.09219

TLDR

AI research agents combine prior knowledge, public sources, and experimental feedback to produce useful results. The Discovery Certification Protocol (DCP) turns claims about these results into executable recovery and feedback tests. Gate 1 validates useful improvement on sealed evaluation. Gate 2 gives matched agents the registered starting information and observed Web content while withholding the target research history. Every valid method reaching the numerical target supplies a recovery witness and triggers the Core veto. DCP Core requires adequate controls, zero observed recoveries, and

arXiv ↗ 中文全文