Speciesist language and nonhuman animal bias in English Masked Language Models

作者:

Highlights:

• We conducted research to examine if language models inherit biases toward animals.

• We used template sentences and sentences extracted from a corpus to test the biases.

• We found that some MLMs associate harmful words with non-human animals.

摘要

•We conducted research to examine if language models inherit biases toward animals.•We used template sentences and sentences extracted from a corpus to test the biases.•We found that some MLMs associate harmful words with non-human animals.

论文关键词:Social bias,Bias evaluation,Masked Language Model,Animal ethics,Speciesism

论文评审过程:Received 13 March 2022, Revised 1 July 2022, Accepted 24 July 2022, Available online 9 August 2022, Version of Record 9 August 2022.

论文官网地址:https://doi.org/10.1016/j.ipm.2022.103050