一句话：提出人机协作需共享知识空间，整合系统文档与隐性专家知识，并以差分测试为案例原型。
软件过程不仅依赖代码、版本历史和日志等有据可查的工件，也依赖团队沟通与隐性专家知识，而这些很少被形式化捕获。作者主张，有效的人机（AI）协作需要共享知识空间，将上述来源整合起来并显式建立连接。论文以差分测试作为动机案例，描述初始原型及其暴露的挑战，并勾勒支持此类协作的动态知识空间愿景。该工作属于立场与愿景型研究，通过原型探索揭示隐性知识与系统知识连接中的问题，为后续构建人机协作知识基础设施提供方向。
Software processes rely not only on well-documented artifacts like code, version histories, and logs, but also on team communications and tacit expertise that are rarely captured formally. We argue that effective human–AI collaboration requires shared knowledge spaces that integrate these sources and make their connections explicit. Using differential testing as a motivating case, we describe our initial prototype and the challenges it revealed, and outline a vision for dynamic knowledge spaces that support such collaboration.
@inproceedings{DearstyneBBD26,
title = {Revealing the Dark Matter: Connecting Tacit and System Knowledge in Human-AI Collaborations},
author = {Katherine R. Dearstyne and Christian Bird and Carmen Badea and Robert DeLine},
booktitle = {ICSE},
year = {2026},
doi = {10.1145/3786582.3786833}
}