Sociotechnical Effects of Machine Translation

TL;DR

The chapter analyzes the sociotechnical effects of machine translation (MT) and large language models (LLMs), emphasizing environmental, social, ethical, and governance dimensions. It advocates a Green AI framework and a triple bottom line approach to evaluate MT systems alongside performance, promoting smaller, energy-aware models and transparent impact reporting. It highlights crisis translation as a high-impact, carefully staged application, offering practical guidelines for rapid deployment, crisis-specific parallel corpora, and phased model development. The discussion also covers data rights, corporate concentration in MT research, end-user literacy, and the potential for MT to empower vulnerable communities when designed with ethics and human-in-the-loop safeguards.

Abstract

While the previous chapters have shown how machine translation (MT) can be useful, in this chapter we discuss some of the side-effects and risks that are associated, and how they might be mitigated. With the move to neural MT and approaches using Large Language Models (LLMs), there is an associated impact on climate change, as the models built by multinational corporations are massive. They are hugely expensive to train, consume large amounts of electricity, and output huge volumes of kgCO2 to boot. However, smaller models which still perform to a high level of quality can be built with much lower carbon footprints, and tuning pre-trained models saves on the requirement to train from scratch. We also discuss the possible detrimental effects of MT on translators and other users. The topics of copyright and ownership of data are discussed, as well as ethical considerations on data and MT use. Finally, we show how if done properly, using MT in crisis scenarios can save lives, and we provide a method of how this might be done.

Figures (2)

• Figure 1: From expediency to quality: how to rapidly deploy LLMs for crisis MT.
• Figure 2: Corpus development process.