LLMs produce racist output when prompted in African American English

Su Lin Blodgett; Zeerak Talat

doi:10.1038/d41586-024-02527-x

LLMs produce racist output when prompted in African American English

Su Lin Blodgett
, Zeerak Talat

Research output: Contribution to journal › Comment/debate › peer-review

Abstract

Large language models (LLMs) are becoming less overtly racist, but respond negatively to text in African American English. Such ‘covert’ racism could harm speakers of this dialect when LLMs are used for decision-making.

Original language	English
Pages (from-to)	40-41
Number of pages	2
Journal	Nature
Volume	633
DOIs	https://doi.org/10.1038/d41586-024-02527-x
Publication status	Published - 28 Aug 2024

Keywords / Materials (for Non-textual outputs)

language
machine learning
society

Access to Document

10.1038/d41586-024-02527-x

Cite this

@article{8da50a4559ab470da600dbf1be41c07b,

title = "LLMs produce racist output when prompted in African American English",

abstract = "Large language models (LLMs) are becoming less overtly racist, but respond negatively to text in African American English. Such {\textquoteleft}covert{\textquoteright} racism could harm speakers of this dialect when LLMs are used for decision-making.",

keywords = "language, machine learning, society",

author = "Blodgett, \{Su Lin\} and Zeerak Talat",

note = "Publisher Copyright: {\textcopyright} Springer Nature Limited 2024.",

year = "2024",

month = aug,

day = "28",

doi = "10.1038/d41586-024-02527-x",

language = "English",

volume = "633",

pages = "40--41",

journal = "Nature",

issn = "0028-0836",

publisher = "Nature Publishing Group",

}

TY - JOUR

T1 - LLMs produce racist output when prompted in African American English

AU - Blodgett, Su Lin

AU - Talat, Zeerak

N1 - Publisher Copyright: © Springer Nature Limited 2024.

PY - 2024/8/28

Y1 - 2024/8/28

N2 - Large language models (LLMs) are becoming less overtly racist, but respond negatively to text in African American English. Such ‘covert’ racism could harm speakers of this dialect when LLMs are used for decision-making.

AB - Large language models (LLMs) are becoming less overtly racist, but respond negatively to text in African American English. Such ‘covert’ racism could harm speakers of this dialect when LLMs are used for decision-making.

KW - language

KW - machine learning

KW - society

UR - https://www.scopus.com/pages/publications/85202523030

U2 - 10.1038/d41586-024-02527-x

DO - 10.1038/d41586-024-02527-x

M3 - Comment/debate

C2 - 39198606

AN - SCOPUS:85202523030

SN - 0028-0836

VL - 633

SP - 40

EP - 41

JO - Nature

JF - Nature

ER -

LLMs produce racist output when prompted in African American English

Abstract

Keywords / Materials (for Non-textual outputs)

Access to Document

Other files and links

Fingerprint

Cite this