Automating Translation

2024-07-22

AI Summary

This study systematically examines the socio-technical impacts of neural machine translation (NMT) and large language models (LLMs), addressing critical issues including environmental cost, translator rights, data ownership, and ethical crisis response. To mitigate these challenges, we propose the first low-carbon MT practice framework, integrating parameter-efficient fine-tuning (PEFT) for carbon reduction, translator-centric collaboration safeguards, and a verifiable MT deployment protocol tailored for humanitarian contexts. We further develop carbon footprint modeling and human-AI collaborative evaluation methodologies. Empirical results demonstrate that lightweight fine-tuning reduces training carbon emissions by over 90%; our ethically grounded deployment protocol constitutes the world’s first standardized framework for humanitarian MT applications, enabling sub-second cross-lingual transmission of critical information in simulated disaster scenarios. The work establishes both theoretical foundations and actionable paradigms for sustainable, responsible MT development.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLP
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search

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.

Problem

• Machine translation's climate impact from large models
• Ethical concerns in data usage and copyright
• Mitigating risks while leveraging MT benefits

Innovation

• Neural MT and LLMs reduce climate impact
• Smaller models lower carbon footprints
• Tuning pre-trained models avoids retraining