AI and audiovisual translation : evaluating the roles of GenAI platforms and prompts in automatic subtitling

Abstract

This chapter examines the potential of Generative AI (GenAI) in automatic subtitling through a comparative case study of fansubbers, ChatGPT, and Google Gemini. Analysing subtitles for a scene from Modern Family, the study evaluates how these sources handle common subtitling challenges like wordplay, multimodal humour, and orality features. The findings reveal that while GenAI models, particularly those capable of processing multimodal information, show promise in addressing some subtitling issues, they still fall short of creative problem-solving and cultural adaptation. The study also explores the impact of different prompting strategies on GenAI performance. The authors discuss practical implications for integrating GenAI into subtitling workflows and emphasise the continued importance of human creativity in audiovisual translation.