Skopos as a Source of Human Creativity in an AI Environment

Abstract

The term ‘creativity’ has been defined and/or used to designate a variety of different concepts in Translation Studies. In recent times, the rise of artificial in­tel­li­gence has further complicated the term’s semantics. Both neural machine-translation systems and large language models are able to produce translations that can be con­sid­ered creative in terms of definitions such as Kußmaul’s (2000: 31), according to which “a creative translation springs from an obligatory modification to the source text and constitutes something that is more or less new and is accepted as more or less appropriate in a (sub-)culture of experts […] at a specific time and with regard to a specific intended purpose.” Where, then, is the scope for human creativity? I shall sug­gest that functionalist approaches such as Vermeer’s skopos theory may provide one answer to this question. My starting point will be a version of Kußmaul’s creativity concept modified to meet the requirements of a skopos-theoretical approach. On this basis, I shall show that some skopoi, in a variety of translation situations, are still quite hard for AI systems to achieve. The examples I shall discuss will be taken from two post-editing courses I held in summer 2023 at FTSK Germersheim.