LT-LiDER — an Interdisciplinary Project — (and Where the Language Services Industry and Its Technology Appear to Be Heading)

It is virtually impossible nowadays to glance at one’s professional communities on social media without noticing yet another debate about artificial intelligence (AI) — especially large language models (LLMs) and (neural) machine translation (N)MT when translators and language services providers are involved. The supporters of LLMs and NMT tend to advocate a bright, ever-cheaper-for-clients future in the language services industry, thanks to the NMT or LLM-related services they or their current companies are either repackaging with minimum effort1 or, much less often, carefully, ethically, and thoughtfully crafting themselves. Across the virtual barricades, the opponents are usually skilled, experienced language specialists (translators, interpreters, subtitlers, terminologists, etc.). Most of them remain unconvinced about the ubiquitous messages proclaiming tech’s added value in virtually all scenarios, after having seen the performance of these technologies in real life on content which matters, and not just on (often questionable, crowdsourced) evaluation datasets. In the middle, there is an important group of hobbyists and knowledge workers who offer language services in languages and/or domains they do not fully (care to) master, or for content which does not carry much risk, and for whom recent AI advances in some language pairs have been genuinely beneficial both in terms of output quality and productivity.