Scholarlink

<p>Knowledge systems rarely collapse under contradiction. They adapt by absorbing it. To explain this dynamic, this article develops <i>epistemic laundering</i> as a theoretical framework. It then shows how Generative AI implements this dynamic on two layers: first, through architectural choices that embed contestable philosophical commitments while concealing that embedding; and second, through the field’s institutional discourse that naturalizes the result as technical achievement.</p><p>Synthesizing Latour’s <i>fabrication of facts</i>, Bourdieu’s <i>misrecognition</i>, Foucault’s <i>truth-regimes</i>, the framework theorizes a recursive mechanism whereby systems designed for knowledge production naturalize their drift into self-stabilizing closure. The mechanism operates through <i>double concealment</i>: systems come to misrecognize their own constructions as reality while concealing this substitution as natural progress. Over time, they become resistant to registering contradiction on its own terms and instead metabolize it.</p><p>Generative AI operationalizes this phenomenon as a sociotechnical infrastructure: contemporary systems operate within closed computational architectures, recombining training distributions while performing the gestures of understanding. AI’s institutional uptake ratifies this machinic performance through discourses surrounding current and future capabilities, including "AGI." Such discourse marks not a technical frontier but the point at which computational criteria are naturalized as the structure of mind itself. What appears as AI’s epistemic ambition is, structurally, the mechanism of its own insulation from correction, exposing how knowledge systems generally persist not by resolving contradiction but by metabolizing it.</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

Epistemic laundering: generative AI and the naturalization of misrecognition

  • 期刊论文

  • 发表日期：

    2026年4月30日

查看全文
  • Theodore Kalaitzidis

摘要

Knowledge systems rarely collapse under contradiction. They adapt by absorbing it. To explain this dynamic, this article develops epistemic laundering as a theoretical framework. It then shows how Generative AI implements this dynamic on two layers: first, through architectural choices that embed contestable philosophical commitments while concealing that embedding; and second, through the field’s institutional discourse that naturalizes the result as technical achievement.

Synthesizing Latour’s fabrication of facts, Bourdieu’s misrecognition, Foucault’s truth-regimes, the framework theorizes a recursive mechanism whereby systems designed for knowledge production naturalize their drift into self-stabilizing closure. The mechanism operates through double concealment: systems come to misrecognize their own constructions as reality while concealing this substitution as natural progress. Over time, they become resistant to registering contradiction on its own terms and instead metabolize it.

Generative AI operationalizes this phenomenon as a sociotechnical infrastructure: contemporary systems operate within closed computational architectures, recombining training distributions while performing the gestures of understanding. AI’s institutional uptake ratifies this machinic performance through discourses surrounding current and future capabilities, including "AGI." Such discourse marks not a technical frontier but the point at which computational criteria are naturalized as the structure of mind itself. What appears as AI’s epistemic ambition is, structurally, the mechanism of its own insulation from correction, exposing how knowledge systems generally persist not by resolving contradiction but by metabolizing it.