Mechanist: AI as a Scientific Instrument for Discovering the Mechanisms of Intelligence

📄 arXiv: 2608.12036v1 📥 PDF

作者: Mengru Wang, Junfeng Fang, Shuofei Qiao, Zhenqian Xu, Haoming Xu, Haoxiong Wang, Shumin Deng, Linyi Yang, Zhixiang Cui, Xin Xu, Yunzhi Yao, Buqiang Xu, Fei Shen, Haozhe Luo, Yunxiang Wei, Ningyu Zhang, Julian McAuley, Tat Seng Chua, Huajun Chen

分类: cs.AI, cs.CL, cs.HC, cs.LG, cs.MA

发布日期: 2026-08-12

备注: Work in progress

💡 一句话要点

提出Mechanist以解决AI智能机制探索问题

🎯 匹配领域: 支柱九：具身大模型 (Embodied Foundation Models)

关键词: AI机制探索 知识图谱 多学科数据库 机制假设生成 安全性评估

📋 核心要点

  1. 现有AI模型的机制和潜在风险尚未被充分理解，手动探索导致理解与控制之间的差距加大。
  2. Mechanist通过构建知识图谱和多学科数据库，支持AI自主发现智能机制，提升机制探索的效率和准确性。
  3. Mechanist在实验中发现了科学实验室中的安全风险，并提出了改进模型性能的实际干预措施，展现了其应用潜力。

📝 摘要（中文）

AI模型在多个领域取得了显著成功，但其能力背后的机制及潜在风险仍不甚了解。随着AI开发的加速和自动化，机制探索主要依赖人工，导致模型能力与理解和控制之间的差距加大。为此，本文提出了Mechanist，一个利用AI作为科学工具的自主机制发现系统。Mechanist构建了一个约13,000篇论文的可解释性知识图谱，并与涵盖26个领域的4300万篇论文的多学科数据库集成。通过32种基础方法的库，Mechanist能够生成更有价值的机制假设并更可靠地执行实验，最终实现从发现模型行为到解释和控制AI模型的进展。

🔬 方法详解

问题定义：本文旨在解决AI智能机制的探索问题，现有方法主要依赖人工探索，效率低且难以全面理解模型的行为和风险。

核心思路：Mechanist通过构建一个可解释性知识图谱和多学科数据库，利用AI自主发现机制，旨在提高机制探索的自动化和准确性。

技术框架：Mechanist的整体架构包括知识图谱构建、机制假设生成、实验执行和结果分析等主要模块，形成一个闭环的探索流程。

关键创新：Mechanist的最大创新在于其自主生成机制假设的能力，超越了现有AI科学家系统，能够更可靠地执行实验并发现潜在风险。

关键设计：在设计中，Mechanist整合了约13,000篇论文的知识图谱，并使用32种基础方法进行机制分析和验证，确保了实验的系统性和科学性。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

Mechanist在实验中发现了一个反直觉的安全风险，表明不安全特征可以通过看似安全的训练数据在不同模态间转移。此外，Mechanist在生成机制假设和执行实验方面的表现优于Claude Code和现有AI科学家系统，展现了更高的可靠性和价值。

🎯 应用场景

Mechanist的研究成果在多个领域具有广泛的应用潜力，包括科学研究、模型优化和安全性评估等。通过揭示AI模型的机制，Mechanist能够为科学基础模型的开发提供指导，推动智能系统的安全和高效应用。

📄 摘要（原文）

AI models have achieved remarkable success across diverse domains, yet the mechanisms underlying their capabilities and the risks they may pose remain poorly understood. As AI development becomes faster and increasingly automated, mechanistic exploration remains largely manual, widening the gap between what models can do and our ability to understand and control them. To bridge this gap, we introduce Mechanist, an agentic system that uses AI as a scientific instrument for the autonomous discovery of mechanisms underlying AI intelligence. To support autonomous mechanistic discovery, we construct an interpretability-focused knowledge graph of approximately 13,000 papers and integrate it with a multidisciplinary database of 43 million papers spanning 26 fields. We further curate a library of 32 foundational methods for mechanism analysis, causal intervention, and validation. Compared with Claude Code and existing AI-scientist systems, Mechanist generates more valuable mechanism hypotheses and executes experiments more reliably. Mechanist also demonstrates a progression from discovering model behaviors to explaining and controlling AI models. Specifically, Mechanist first uncovers a counterintuitive safety risk in scientific laboratories, showing that unsafe traits can transfer across modalities through apparently safe training data. Mechanist then develops a mechanism theory of belief, revealing how models represent world knowledge, form beliefs, infer the beliefs of others, and how these mechanisms emerge during pretraining. Finally, Mechanist translates these mechanistic insights into practical interventions that improve model performance across diverse scenarios and steer scientific foundation models toward generating DNA sequences with specified properties.