Abstract：This talk begins by examining the state of machine translation technology at the time of the establishment of AAMT, as well as the relationship between machine translation and professional human translators. It then discusses how successive technological paradigm shifts have transformed the role of translators over the past 35 years.
From early rule-based machine translation and translation memory systems, through statistical machine translation and neural machine translation, to today's translation based on large language models (LLMs) and emerging agentic AI frameworks, the roles of engineers, translators, and national institutions have undergone significant and continuous change. Throughout this evolution, the functions of organizations such as AAMT and IAMT—founded to foster close communication among researchers, translators, and public institutions—as well as the MT Summit conferences they have organized, have also evolved. The skepticism that many translators initially expressed toward machine translation is now becoming increasingly visible once again with the emergence of machine translation within agentic AI environments. As machine translation continues to reshape translation workflows and transform business models within the translation industry, this talk looks back on the major developments and transitions of the past 35 years.

Abstract：This presentation examines translation in Taiwan in the age of artificial intelligence by taking government translation outsourcing as a starting point for discussing broader changes in the translation profession and translator education. Moving beyond reductive narratives that frame AI as simply replacing human translation, the presentation begins with public-sector outsourcing as evidence of the continuing institutional demand for professional translation in Taiwan. Recent translation-related procurement requests on the government procurement website show that public-sector translation demand extends across literature, public infrastructure, international exchange, publishing, human rights reporting, manuscript and data processing, legal forums, cybersecurity, and museum collections. These projects also suggest that government translation outsourcing often involves a broad range of language services, including written translation, interpretation, editing, documentation support, event communication, and metadata translation.
These procurement examples do not simply demonstrate the persistence of translation demand; they also point to the kinds of institutional, public-facing, and domain-specific texts for which AI-assisted translation must be carefully evaluated. In this sense, government outsourcing provides a concrete context for examining how machine translation, large language models, and post-editing workflows may be integrated into professional translation practice, and what forms of human judgment remain necessary when such tools are used. The issue, therefore, is not whether AI will replace translators, but how translators can work critically and responsibly with AI-assisted tools in contexts where accuracy, consistency, terminology, accountability, and communicative purpose are essential.
The presentation then turns to translator education in Taiwan and asks how training programs can respond to these changing professional conditions. It argues that translator education should move beyond general awareness of translation technology toward the systematic development of critical, ethical, and strategic competence. Students need to be prepared not only as users of AI-assisted translation tools, but also as language professionals capable of assessing when such tools are appropriate, how machine-generated outputs should be revised, and what forms of human expertise remain indispensable. By connecting government outsourcing, AI-assisted translation, and translator education, the presentation highlights the need to rethink translation training in relation to both market demand and public communication needs, and concludes by considering future directions for human-machine collaboration, public-sector translation quality, and the evolving role of translators in the AI era.

Abstract：Generative AI is beginning to reshape translation as part of a broader range of language and knowledge tasks. This talk focuses mainly on the challenges that accompany this development: the opacity of foundation models, uneven linguistic and cultural coverage, and questions of trust, provenance, and technological sovereignty. Drawing on the activities of LLM-jp and the NII Large Language Model R&D Center, I will discuss the value of open foundation models as transparent research infrastructure. I will also consider how human expertise can be cultivated and passed on as AI assumes a growing role in intellectual work.

Abstract：SEA-LION is developed by the AI Products team at AI Singapore (AISG), a national AI programme under NRF, building Singapore's deep AI capabilities across research, products and talent.
In this talk, Jian Gang and Jeongmi will introduce SEA-LION, a family of open-source models trained to support Southeast Asian (SEA) languages, cultures and contexts, SEA-HELM, a comprehensive evaluation suite for diagnosing LLMs and their understanding of SEA safety, languages and cultures, as well as various AI tools that support the SEA region, including for machine translation.

Abstract：This session introduces how the AI simultaneous interpretation service "Live Translation" works, along with real-world use cases. No dedicated hardware or app installation is required: participants simply scan a QR code, and the translation appears on their own device in their own language. Accuracy can be improved further by registering proper nouns and technical terminology in advance. The session includes a live demonstration, together with a look at how the service is used at international conferences, corporate events, and in-house gatherings. We hope it offers a useful additional option to consider for multilingual support.