Scholarlink

<p>Extended cognition theory holds that humans tend to rely on external materials, such as notebooks, calculators, and software, to support their cognitive processes when building things. But humans are mortal, while their creations live longer. So when the people who created and understood a system retire or die, the artifacts persevere, still functional but severed from the human practices that made them intelligible. Engineers call this epistemic debt. The legacy COBOL case study is a good example. Industry estimates suggest that at least 220&#xa0;billion lines of COBOL code still process most global financial transactions today, but the engineers who wrote them have mostly passed away. Generative AI might be accelerating this problem. Vibe-coded artefacts enter infrastructure production already opaque because the human initiator who created them never built a mental model of the logic. So the window between software code creation and incomprehensibility is collapsing from generations to sprints. Modern governance presupposes that someone at some point understood what was built. But epistemic debt undermines this. Using distributed cognition theory and engineering research, we argue that institutions should monitor <i>comprehension half-life</i>, which is the rate at which the understanding of a system decays after its creation, and invest in epistemic infrastructure that preserves intelligibility across generations. Extended minds are mortal. Artifacts created by them are not. Governing this asymmetry is an emerging intergenerational challenge.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Mortality of Distributed Understanding

  • 期刊论文

  • 发表日期：

    2026年5月1日

查看全文
  • Alexander C. Ozoani,
  • Samuel O. Oyefusi

摘要

Extended cognition theory holds that humans tend to rely on external materials, such as notebooks, calculators, and software, to support their cognitive processes when building things. But humans are mortal, while their creations live longer. So when the people who created and understood a system retire or die, the artifacts persevere, still functional but severed from the human practices that made them intelligible. Engineers call this epistemic debt. The legacy COBOL case study is a good example. Industry estimates suggest that at least 220 billion lines of COBOL code still process most global financial transactions today, but the engineers who wrote them have mostly passed away. Generative AI might be accelerating this problem. Vibe-coded artefacts enter infrastructure production already opaque because the human initiator who created them never built a mental model of the logic. So the window between software code creation and incomprehensibility is collapsing from generations to sprints. Modern governance presupposes that someone at some point understood what was built. But epistemic debt undermines this. Using distributed cognition theory and engineering research, we argue that institutions should monitor comprehension half-life, which is the rate at which the understanding of a system decays after its creation, and invest in epistemic infrastructure that preserves intelligibility across generations. Extended minds are mortal. Artifacts created by them are not. Governing this asymmetry is an emerging intergenerational challenge.