arXiv cs.CL· Iwan Williams, Ninell Oldenburg, Ruchira Dhar, Joshua Hatherley, Constanza Fierro, Nina Rajcic, Sandrine R. Schiller, Filippos Stamatiou, Anders S{\o}gaard·· 10 小时前AI 评分26
机械可解释性需要哲学:一篇 NeurIPS 2026 立场论文
Mechanistic Interpretability Needs Philosophy
AI 导读
一篇被 NeurIPS 2026 接收为 Spotlight 的立场论文提出,机械可解释性(MI)研究需要哲学作为持续伙伴,以厘清概念、改进方法并应对解读 AI 系统时的认识论与伦理复杂性。文章以 MI 文献中的三个开放问题为例,说明哲学框架能为 MI 研究带来的价值,并勾勒出深化跨学科对话的路径。
正文 · 原文
Abstract:Mechanistic interpretability (MI) aims to explain how trained neural networks work by uncovering their underlying mechanisms. As the field grows in influence, it is increasingly important to examine not just models themselves, but the assumptions, concepts and explanatory strategies implicit in MI research. We argue that mechanistic interpretability needs philosophy as an ongoing partner in clarifying its concepts, refining its methods, and navigating the epistemic and ethical complexities of interpreting AI systems. There is significant unrealised potential for progress in MI to be gained through deeper engagement with philosophers and philosophical frameworks. Taking three open problems from the MI literature as examples, this paper illustrates the value philosophy can add to MI research, and outlines a path toward deeper interdisciplinary dialogue.
| Comments: | Accepted at NeurIPS 2026 (Spotlight, Position Paper Track) |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2506.18852 [cs.CL] |
| (or arXiv:2506.18852v3 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2506.18852 arXiv-issued DOI via DataCite |
Submission history
From: Iwan Williams [view email]
[v1]
Mon, 23 Jun 2025 17:13:30 UTC (262 KB)
[v2]
Tue, 19 May 2026 12:40:52 UTC (257 KB)
[v3]
Thu, 19 Nov 2026 12:26:25 UTC (240 KB)
来源:arXiv cs.CL · arxiv.org