The Story is Not the Science: Execution-Grounded Evaluation of Mechanistic Interpretability Research
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866915810759409664 |
|---|---|
| author | Bai, Xiaoyan Baumgartner, Alexander Sun, Haojia Holtzman, Ari Tan, Chenhao |
| author_facet | Bai, Xiaoyan Baumgartner, Alexander Sun, Haojia Holtzman, Ari Tan, Chenhao |
| contents | Reproducibility crises across sciences highlight the limitations of the paper-centric review system in assessing the rigor and reproducibility of research. AI agents that autonomously design and generate large volumes of research outputs exacerbate these challenges. In this work, we address the growing challenges of scalability and rigor by flipping the dynamic and developing AI agents as research evaluators. We propose the first execution-grounded evaluation framework that verifies research beyond narrative review by examining code and data alongside the paper. We use mechanistic interpretability research as a testbed, build standardized research output, and develop MechEvalAgent, an automated evaluation framework that assesses the coherence of the experimental process, the reproducibility of results, and the generalizability of findings. We show that our framework achieves above 80% agreement with human judges, identifies substantial methodological problems, and surfaces 51 additional issues that human reviewers miss. Our work demonstrates the potential of AI agents to transform research evaluation and pave the way for rigorous scientific practices. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_18458 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | The Story is Not the Science: Execution-Grounded Evaluation of Mechanistic Interpretability Research Bai, Xiaoyan Baumgartner, Alexander Sun, Haojia Holtzman, Ari Tan, Chenhao Computers and Society Artificial Intelligence Computation and Language Machine Learning Reproducibility crises across sciences highlight the limitations of the paper-centric review system in assessing the rigor and reproducibility of research. AI agents that autonomously design and generate large volumes of research outputs exacerbate these challenges. In this work, we address the growing challenges of scalability and rigor by flipping the dynamic and developing AI agents as research evaluators. We propose the first execution-grounded evaluation framework that verifies research beyond narrative review by examining code and data alongside the paper. We use mechanistic interpretability research as a testbed, build standardized research output, and develop MechEvalAgent, an automated evaluation framework that assesses the coherence of the experimental process, the reproducibility of results, and the generalizability of findings. We show that our framework achieves above 80% agreement with human judges, identifies substantial methodological problems, and surfaces 51 additional issues that human reviewers miss. Our work demonstrates the potential of AI agents to transform research evaluation and pave the way for rigorous scientific practices. |
| title | The Story is Not the Science: Execution-Grounded Evaluation of Mechanistic Interpretability Research |
| topic | Computers and Society Artificial Intelligence Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2602.18458 |