다문화 시대의 한국어 의사소통과 인공지능 -차별 배제와 공정 지향에 대한 인식을 중심으로-

초록

We examine a number of Artificial Intelligence(AI) related issues in the context of Korean communication in the age of multiculturalism. We highlight the fact that thanks to the specifics of current AI technology, Korean communication with AI could worsen the already serious social bias and discrimination problem. We point out that pre-existing social bias and discrimination tendency could be reflected in the training data, even without explicit algorithmic bias. We also claim that historical injustice could also influence in the current and future AI decision-making given this training-data dependency. We argue that we need to take into account pluralistic attitude towards various frames of cultural interpretation as well as danger of stereotype and contextuality in order to trust Korean communication by AI, for example in the context of interviewing process.

키워드