본문 바로가기
  • Home

AI Bias Analysis Model Based on Prototype Theory

  • 탈경계인문학Trans-Humanities
  • 2026, 19(2), pp.116~142
  • DOI : 10.22901/trans.2026.19.2.116
  • Publisher : Ewha Institute for the Humanities: EIH
  • Research Area : Humanities > Other Humanities
  • Received : September 10, 2026
  • Accepted : October 15, 2026
  • Published : October 31, 2026

In-Ryeong CHOI 1

1서울대학교

Accredited

ABSTRACT

The definition of ‘AI bias’ is ambiguous. Although it has been studied from statistical, cognitive, and socio-historical perspectives, such research has proceeded without a theoretical foundation of what bias is. Recognizing this problem, we redefine the concept of bias based on Lakoff’s prototype theory, in order to explain the cognitive and linguistic mechanisms underlying bias. We also propose a new analytical model for bias in artificial intelligence. After critically examining Davat’s definition of “bias as a deviation from the norm”, our study concludes that the degree of bias depends on how far it deviates from the central prototype of a category. It is also demonstrated that, in a specific socio-cultural context, the degree of bias is inversely proportional to prototypicality, indicating that bias is not a matter of deviation from neutral norm but of selecting a prototype.

Journal Copyright Policy

No CCL information provided

Citation status

* References for papers published after 2025 are currently being built.

This paper was written with support from the National Research Foundation of Korea.