Introduction: Toward human-centred AI in translation studies

Abstract

Artificial intelligence (AI) is now woven into translation in ways that were scarcely imaginable a decade ago. Neural machine translation (NMT) built on the Transformer architecture (Vaswani et al., 2017) has altered expectations of speed, coverage and quality, while increasingly multimodal, massively multilingual systems promise “any-to-any” translation across text and speech. Yet if translation studies is to harness these advances responsibly, we need to place people (translators, interpreters, localizers, educators, communities, and end-users) at the center of system design, policy, and evaluation. This is the core proposition of human-centered AI (HCAI): design AI that amplifies human agency, mastery, and accountability, rather than displacing them (Shneiderman, 2022, for HCAI in translation studies see Jiménez-Crespo, 2025).

HCAI is not a vague ideal. In human–computer interaction (HCI), concrete guidance exists for how AI systems should behave in practice (e.g., the “Guidelines for Human-AI Interaction,” which articulate principles like keeping users in control, supporting efficient correction, and exposing system confidence). These guidelines are directly relevant to translation tools, from interactive MT to quality estimation dashboards and post-editing environments. At the same time, broader regulatory frameworks now impose obligations on providers and deployers of AI. In the European Union, for example, the AI Act entered into force on 1 August 2024 with phased applicability through 2025–2027, including near-term prohibitions and obligations for general-purpose AI and later rules for high-risk systems – developments with clear implications for translation workflows and vendors.

This issue on Bridge takes up the challenge of human-centered AI in translation by foregrounding translators’ expertise and lived practices, domain-specific quality, equity for under-resourced languages and modalities, and educational and regulatory transitions. Below is a sketch of the technological backdrop, and a map of key human-centered concerns.