Simultaneous Interpretation Generated by Artificial Intelligence. An Arabic-Spanish Case Study

Abstract Few areas have been unaffected by the emergence of Artificial Intelligence (AI) today. Language interpretation is no exception. Given that AI is based on large linguistic models and the use of increasingly sophisticated algorithms, its applications in interpretation are advancing at a previously unimaginable pace. However, in times like these, which are largely conducive to the implementation of AI in all areas, it is more necessary than ever to study the consequences of this technology and how it can affect the quality of interpreting. While its positive contribution as an assistant in work processes is undeniable, there are many challenges in terms of professional sustainability, ethics and product quality that are emerging in the present and immediate future. This contribution critically analyses a real case of AI-Voice Translation (AI-VT) used as a simultaneous interpreting (SI) system. We are aware that this practice is no longer an exception in the professional market in our local environment. The case study was a real assignment consisting of a conference in Arabic held at the Faculty of Translation and Interpreting on 11 April 2025. The conference, which lasted an hour and a half, focused on Algerian theatre with an analysis of the sociocultural features of the play ‘hafila tasir’ (the bus continues its way). Two ‘interpretation’ systems were used. On the one hand, AI-VT was provided by a company based in Granada (Spain) through the service it has contracted on the LiveVoice platform. On the other hand, the interpretation was carried out by two students from the Master's Degree in Conference Interpreting at the University of Granada, who had thoroughly researched this work. The event was attended by a small audience that could hear both the human interpretation (HI) and the AI-VT. It should be noted that the assignment was demanding due to the speed of the speaker and the use of video clips in Arabic dialect. Based on the recording and transcription of the HI and the AI-VT, this work aims to offer a holistic evaluation of the AI-VT quality comparing it with HI considering the requirements demanded in today's SI market. The first general results indicate that AI-VT, used as a professional SI system, suffers from a significant loss of quality that affects all the parameters analysed, which makes its use for this purpose inadvisable.