The human factor is not lost, only transformed

In early October, the BME Centre for Modern Languages (INYK) organised the Artificial Intelligence for Translation and Interpreting (AI for T/I) conference, with the support of the EELISA alliance and in cooperation with ZHAW in Zurich and UPM in Madrid. The two-day English-language event was also available online, with the organisers registering 274 views on the first day and 137 on the second.

The central theme was the exploration of the relationship between artificial intelligence (AI) and language mediation. Researchers, educators, and market players from the fields of artificial intelligence, engineering sciences, and language mediation presented and discussed new research findings and industry trends. In the spirit of interdisciplinary knowledge sharing, 

the conference provided a forum for representatives of various scientific fields and the language industry, as well as international and Hungarian students.

On the first day, following a welcome speech by Mária Szalmáné Csete, Dean of the Faculty of Economics and Social Sciences, Csaba Hegedűs, a doctoral student at the Department of Telecommunications and Artificial Intelligence, delivered a presentation entitled Fundamentals of Large Language Models, in which he presented the operating principles and potential applications of large language models (LLMs) in education and translation technology.

Csaba Hegedűs

The practical side of technological developments was presented by Márton Jánosy, CEO of Edimart. In his lecture entitled Where Packaged AI Tools Fall Short — and What We Built Instead, he used self-developed AI solutions to illustrate how to overcome the limitations of commercially available tools. The afternoon session featured parallel workshops: Levente Péter Nagy (reviser and translation technologist, MKIFK) presented how ChatGPT can be used to support the work of translators in his workshop entitled "ChatGPT as a Translation Engine and Translator Assistant". At the same time, Pablo Calleja and Patricia Martín-Chozas (UPM) explored the linguistic foundations of generative AI in their workshop entitled "Linguistics in the Era of LLMs".

Márton Jánosy

Balázs Kis, co-founder of MemoQ, delivered a presentation entitled Quo vadis, Language Professions? Some Best-Case Scenarios, examining the future of language mediation as a profession in the age of artificial intelligence. The day concluded with a roundtable discussion entitled Trainers Meet Professionals: AI for T/I Training – Human in the Loop (Translation). Industry and university representatives, including Krisztián Király (HUNNECT), Anna Martin (DGT), Edina Robin (ELTE), Mattia Turra (ZHAW), Pablo Calleja and Patricia Martín-Chozas, shared their experiences of combining translation training and technological innovation.

The second day was opened by Gergely Vígh, Director of International Study Programmes at BME, followed by a presentation by Franz Pöchhacker (University of Vienna) entitled Automated Interpreting: The Challenge for Research, which analysed the research challenges and ethical issues of machine interpreting. He was followed by Michaela Albl-Mikasa (ZHAW) with her presentation entitled AI Has Intelligence and Creativity, Humans Have Intuition, which emphasised the role of human intuition in AI-supported processes. The morning session was closed by Mihály Héder, Head of the Department of Philosophy and History of Science at BME, delivering his presentation entitled Artificial Intelligence vs. Human Identity.

The afternoon programme consisted of parallel sessions: Márta Fischer, director, and Krisztina Szabó-Horváth, translation project manager at INYK, presented a case study (Challenging the 'AI Can Do It All' Myth) on the creation of BME's internal translation office and terminology database, while Anne Catherine Gieshoff and Mattia Turra (ZHAW) presented the application of AI-generated interactive dialogues in interpreter training in their "IT-COMMIT" workshop.

Márta Fischer

The second round table discussion, entitled "Trainers Meet Professionals: AI for T/I Training – Human in the Loop (Interpreting)", took place at the end of the day. Michaela Albl-Mikasa, Franz Pöchhacker, Anne Catherine Gieshoff, Márta Seresi (ELTE) and Veronika Vaspál (MFTE) discussed issues concerning the balance between AI and human expertise.
Special attention was given to the collaboration of 16 students from five universities of the alliance (BME, ITU in Istanbul, UPB in Bucharest, UPM in Madrid and FAU in Erlangen-Nuremberg) within the framework of EELISA. The conference concluded with a workshop presenting arguments for and against human factors in AI-assisted language mediation, reflecting on the topics covered in the preceding presentations.

The initiative was in line with the mission of the EELISA community "AI for Mediation between Languages and Cultures", founded by researchers at the Centre for Modern Languages and supported three Sustainable Development Goals (SDGs):

•    SDG 4 (Quality Education) by making high-quality AI-based training accessible;
•    SDG 9 (Industry, Innovation and Infrastructure) by encouraging academic and industrial collaborations;
•    and SDG 17 (Partnerships for the Goals) by interdisciplinary networking and knowledge sharing.

The organisers intended the event to be the first in a series, with further events planned by the community.

In addition to dispelling misconceptions about translation and interpreting, the conference made it clear that, in the age of artificial intelligence, the human factor is not being pushed into the background, but is becoming indispensable in a new role and with new competencies –  posing a serious challenge for educational institutions.

Rector's Office, Communications Directorate