METM25 presentation

Synthetic-text editing: two exploratory case studies

Michael Farrell, Mortara, Italy

This presentation explores how generative artificial intelligence, or GenAI, is changing the way authors create texts in a foreign language. With tools like GPT-4, authors are now sidestepping traditional translation methods, opting instead to write directly in their second language. This shift is transforming the role of professional editors, who are moving from correcting translation and non-native errors to refining AI-assisted texts.

We will discuss the findings from a study involving two native Italian speakers who used GPT-4 to craft articles about art education in English. These authors then self-edited their AI-generated drafts before passing them on to professional editors for further enhancement. Both the authors and editors shared their experiences – what worked well and what did not – and described the nature of their interactions throughout the process.

The findings reveal that, while GenAI can improve grammatical accuracy in English, it also introduces anomalies like redundancy and repeated phrases, which require careful review by editors. Examples of these anomalies will be provided. Despite these issues, the authors praised the effectiveness of using GenAI for writing and they plan to continue using it.

The study highlights that although GenAI tools are beneficial, human editors remain essential to produce polished, publishable content. Furthermore, professional translators and editors, with their linguistic expertise, are probably well-equipped to provide synthetic-text editing services. The presentation emphasizes the importance of training both writers and editors to work effectively with these tools in anticipation of a future where GenAI plays a central role in multilingual writing. It also examines the key training needs that arise as a result.

Further reading

Farrell, Michael. 2025. Editing synthetic text from generative artificial intelligence: two exploratory case studies. Proceedings of the 46th Conference Translating and the Computer, Luxembourg, November 18-20, 2024.

About the presenter

Michael Farrell is a translator; researcher at the International Center for Research on Collaborative Translation; and lecturer in localization, post-editing, machine translation, artificial intelligence, and computer tools for translators at IULM University, Milan. He is the developer of IntelliWebSearch and a member of AITI, the European Association for Machine Translation, and MET’s Council.